Specific disulfide cleavage is required for ubiquitin conjugation and degradation of lysozyme.

Both ubiquitin conjugation and ubiquitin-dependent degradation of chicken egg white lysozyme in a reticulocyte lysate depend on the presence of a reducing agent. We present evidence that the reduction of a specific disulfide bond, namely that at Cys6-Cys127, facilitates ubiquitination and is a prerequisite to the formation of a multiubiquitin chain on one of at least four chain initiation sites on lysozyme. The Cys6-Cys127 disulfide bond in lysozyme can be specifically reduced, and the modified protein can be isolated after carboxymethylation of the 2 resulting cysteines. This modified lysozyme no longer requires the presence of a reducing agent for ubiquitin conjugation and degradation. Inhibition of ubiquitination by the dipeptide Lys-Ala revealed that this modified lysozyme, like the unmodified protein, is recognized via the binding of the ubiquitin protein ligase, E3, to the substrate's N-terminal lysyl residue. Both the rate and the extent of ubiquitin-lysozyme conjugation, however, are significantly higher with this modified substrate. Likewise, ubiquitin-dependent degradation of 6,127-reduced/carboxymethylated lysozyme was 2-4-fold faster than degradation of the unmodified counterpart. These results are consistent with an interpretation that the modified lysozyme mimics an intermediate formed at the rate-limiting step of the degradation of lysozyme in the reticulocyte lysate. Reduction of the Cys6-Cys127 disulfide bond is expected to unhinge the N-terminal region of lysozyme, and we propose that the recognition of this otherwise stable protein by the ubiquitin pathway is due to facilitated binding of E3 that results from such a conformational transition.

Both ubiquitin conjugation and ubiquitin-dependent degradation of chicken egg white lysozyme in a reticulocyte lysate depend on the presence of a reducing agent. We present evidence that the reduction of a specific disulfide bond, namely that at Cys'-Cy~'~~, facilitates ubiquitination and is a prerequisite to the formation of a multiubiquitin chain on one of at least four chain initiation sites on lysozyme. T h e C y s ' -C y~'~~ disulfide bond in lysozyme can be specifically reduced, and the modified protein can be isolated after carboxymethylation of the 2 resulting cysteines. This modified lysozyme no longer requires the presence of a reducing agent for ubiquitin conjugation and degradation. Inhibition of ubiquitination by the dipeptide Lys-Ala revealed that this modified lysozyme, like the unmodified protein, is recognized via the binding of the ubiquitin protein ligase, E3, to the substrate's Nterminal lysyl residue. Both the rate and the extent of ubiquitin-lysozyme conjugation, however, are significantly higher with this modified substrate. Likewise, ubiquitin-dependent degradation of 6,127-reducedl carboxymethylated lysozyme was 2-4-fold faster than degradation of the unmodified counterpart. These results are consistent with an interpretation that the modified lysozyme mimics an intermediate formed at the rate-limiting step of the degradation of lysozyme in the reticulocyte lysate. Reduction of t h e C y~~-C y s '~' disulfide bond is expected to unhinge the N-terminal region of lysozyme, and we propose that the recognition of this otherwise stable protein by the ubiquitin pathway is due to facilitated binding of E3 that results from such a conformational transition.
Genetic and biochemical studies have shown that, in eukaryotes, cellular abnormal and damaged proteins are degraded by a ubiquitin-mediated pathway. In this pathway, ubiquitin, a 76-amino acid polypeptide, is linked through its C terminus to form an amide (formally, an isopeptide) bond with a lysyl t-amino group on the substrate protein (reviewed in Refs. 1 and 2). How proteins are targeted for this covalent modification is fundamental to the issue of selective intracel-* This work was supported by United States Public Health Service Grants GM37666 (to R. E. C.) and GM35803 (to V. C.). Purchase of the Jasco 5-600 spectropolarimeter a t UCLA was made possible by National Institutes of Health Grant S10 RR04921. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
8 To whom correspondence should be addressed. lular protein turnover. Moreover, the ubiquitin system further discriminates among potential substrates in that not all ubiquitin conjugation targets are destined for degradation. Understanding the substrate selection process will require a description of those structural features recognized by the ubiquitination enzymes.
For one pathway of ubiquitin-dependent degradation, several studies have pointed to the importance of the N-terminal region of proteins in the substrate recognition step. Bachmair et al. (3), by the use of Escherichia coli P-galatosidase derivatives containing specific N-terminal extensions, showed that the N-terminal residue in these proteins provided a signal for their recognition by the ubiquitin degradation pathway in the yeast Saccharomyces cereuisiae. By variation of the N-terminal residue, the efficacy of each of the 20 common amino acids as the recognition signal was ranked in yeast and, more recently, in rabbit reticulocyte lysate (3,4). The same Nterminal extensions have been shown to confer recognition when fused with the otherwise stable dihydrofolate reductase (5). Involvement of the N-terminus as a recognition signal also has been established in work by Hershko and his colleagues (6, 7), who demonstrated an N-terminal amino acid dependence for the binding of substrates to the ubiquitin protein ligase (also called E3). E3, together with ubiquitin carrier proteins (also called E2 or ubiquitin conjugation enzymes), was required for ubiquitination in vitro of certain proteins such as a-lactalbumin, P-lactoglobulin, and ribonuclease A derivatives.
Notwithstanding the substantial evidence implicating the N-terminal amino acid as a recognition determinant, other features also must be involved in substrate selection. Whereas native ribonuclease A is not ubiquitinated despite having a permissive or "destabilizing" N-terminal residue, a variety of chemically modified forms of the protein are excellent substrates (8,9). Recognition of these derivatives is due not to specific interactions with the damaged or modified amino acids but is a consequence of the unfolding that accompanies the covalent modifications (9). Similarly, in addition to a permissive N-terminal amino acid, ubiquitination of P-galactosidase and dihydrofolate reductase requires a 33-45-residue N-terminal extension that is thought to be unstructured and relatively flexible (5, 10).
Hen (chicken) egg white lysozyme has been used extensively as a substrate to study the ubiquitin-dependent proteolytic pathway in vitro. The recognition of this substrate protein has been shown to require the binding of the ubiquitin protein ligase, E3, to the N-terminal lysine residue in lysozyme (6).
Thus, lysozyme appears to be ubiquitinated via the same pathway as the ribonuclease derivatives and fusion protein substrates discussed above. However, whereas some aspect of "unfoldedness" is involved in the selection of these latter proteins for ubiquitination, native lysozyme has been employed routinely as an effective substrate. In the present study, we provide evidence that native lysozyme in fact is not ubiquitinated, but that, under conventional ubiquitin conjugation and degradation assay conditions, a small fraction of the protein is specifically reduced at one of the four disulfide bonds ( C y~~-C y s '~~) . T h i s is the first step in the recognition of lysozyme for degradation by the ubiquitin pathway in reticulocyte extracts in a process that leads to the formation of a multiubiquitin chain. The ubiquitin moieties in this chain are joined to each other via an isopeptide bond that is formed between the C-terminal Gly76 of one ubiquitin and Lys4' of an adjoining ubiquitin. This structure previously was found to target the degradation of ,&galactosidase (lo), presumably by serving as a docking site for the ubiquitin-dependent protease. Our observations regarding lysozyme ubiquitination and the properties of a three-disulfide lysozyme derivative offer, for one class of E3-dependent substrates, the opportunity to explore the reaction pathway of ubiquitin-mediated proteolysis with a structurally defined protein.

EXPERIMENTAL PROCEDURES
Materials-'ZsII-Labeled proteins were prepared by radioiodination with chloramine T (11) and carrier-free NalZ5I from Amersham Radiochemicals. The dipeptide Lys-Ala, bovine ubiquitin, and hen (chicken) egg white lysozyme (3 X crystallized, grade I) were purchased from Sigma. Ubiquitin was purified further by cation-exchange chromatography using a Pharmacia LKB Biotechnology Inc. Mono S column and fast protein liquid chromatography system. [ l e~c y l -~H ] Ubiquitin was produced in E. coli harboring a plasmid encoding yeast ubiquitin and grown with ["Hlleucine' and was provided by J. Setsuda (UCLA). Ubiquitin-depleted reticulocyte lysate (Fraction 11) was prepared as described previously (12,13) from washed rabbit reticulocytes purchased from Green Hectares (Oregon, WI) except that lysates contained 0.1 mM DTT.' Comparable activity and stability of the lysates were found whether 0.1 mM or the more usual 1 mM DTT was used for storage. Purified ring-necked pheasant lysozyme was generously provided by Drs. E. Prager and A. C. Wilson of the University of California a t Berkeley. Prestained molecular weight markers were from Bethesda Research Laboratories.
Synthesis of Ubiquitin Variants-The chemical alterations in the ubiquitin variants are shown in Fig. 1. Ub-C48 differs from wild-type ubiquitin in having a cysteine a t position 48 instead of a lysine. This variant protein was obtained by expression of a mutated ubiquitin gene in E. coli AR58 cells and was isolated as described previously (13). Reductive methylation and S-aminoethylation of Ub-C48 were done as reported (13) and are described in more detail below. Purified and lyophilized Ub-C48 was dissolved to 1 mg/ml in 8 M urea, 1 mM DTT, and 0.1 M Na+-Hepes, pH 7.0. Sodium cyanoborohydride and formaldehyde were then added sequentially to final concentrations of 20 and 12 mM, respectively. After gentle stirring for 18 h a t room temperature, further additions of the two reagents were made to double their concentrations, and the reaction was continued for an additional hour. The solution was then dialyzed against six changes of 20 volumes of 2 mM DTT and lyophilized. The fraction of unblocked amino groups in the MeUb-C48 preparation was less than 4%, as judged by its reaction with fluorescamine (14). T o convert MeUb-C48 to MeUb-AEtC48 (13), 100 mg of lyophilized MeUb-C48 was dissolved in 30 ml of water, and the sulfhydryl content of this protein solution was determined by its reaction with 5,5'-dithiobis(2nitrobenzoic acid) (15). The solution was then adjusted to 100 ml by the addition of N-ethylmorpholine acetate, pH 8. The abbreviations used are: DTT, dithiothreitol; Hepes, 4-(2-hydroxyethy1)-1-piperazineethanesulfonic acid rcm, reduced and carboxymethylated; SDS, sodium dodecyl sulfate; HPLC, high performance liquid chromatography; Me, methyl; AEt, aminoethyl. stirred gently a t 50 "C. Progress of the reaction was monitored with the sulfhydryl group assay. In general, the reaction was complete within 4 h, and the sample was then dialyzed against water and lyophilized. The products were separated from unreacted protein by cation-exchange chromatography as follows. The proteins were dissolved in 25 mM ammonium acetate, pH 4.5, and loaded onto a 100ml column of carboxymethylcellulose (CM52, from Whatman) equilibrated in the same buffer. After washing with 200 ml of the pH 4.5 buffer, proteins were eluted with a 500-ml gradient of 25 mM ammonium acetate, p H 4.5-7.0. MeUb-C48 eluted at pH 5.2 and MeUb-AEtC48 at pH 5.7, and they were identified by their distinctive conjugation patterns with a (%galactosidase derivative (10). Each peak was pooled and rechromatographed on an LKB SP-5PW column, using the same pH gradient, to obtain the pure proteins. A yield of 30 mg of Ub-AEtC48 was obtained from 100 mg of Ub-C48. The sample was dialyzed against water and stored as a lyophilized powder. MeUb-MeAEtC48 was obtained by reductive methylation of MeUb-AEtC48 using the same procedure for the derivatization of Ub-C48.
Conjugation and Degradation Assays-The conjugation of ubiquitin or its variants to '2sII-lysozyme or to 12611-6,127-rcm-lysozyme in the ubiquitin-depleted reticulocyte lysate was done a t 37 "C for 30 min in a 25-pl reaction containing 50 mM Tris.HCl, pH 7.5, 2 mM DTT, 5 mM MgC12, 2 mM ATP, and an ATP-regenerating system (10 mM creatine phosphate and 0.1 mg/ml creatine phosphokinase) (16). Either no ubiquitin or 50 PM ubiquitin (or one of its variants) was also added, and reactions were stopped by the addition of SDS-gel sample buffer. Alternatively, assays with "51-ubiquitin (2.5 pM, 5 X lo5 cpm in 20 pl) and in a 50 mM Na+-Hepes buffer, pH 7.2, were done as described (9) and yielded similar results. Products were analyzed by gel electrophoresis (17) and autoradiography, using an SDS-polyacrylamide (12%) gel. Assays without DTT supplementation generally contained 0.01 mM DTT, which was introduced with the lysate. Otherwise, complete DTT depletion was achieved by rapid gel filtration of the lysate as described (9). The degradation of '"Ilysozyme or '2sI-6,127-rcm-lysozyme was measured under the same reaction conditions, except that a 50-pl aliquot from a 300-p1 reaction mixture was withdrawn every 30 min, mixed with 0.5 ml of 10% trichloroacetic acid and centrifuged, and the radioactivity in the supernatant was measured with a y-counter. The percent degradation was calculated from the fraction of the total counts converted to an acid-soluble form. The ubiquitin-dependent component of the degradation was determined by subtraction of values obtained when ubiquitin was omitted.
Preparation of 6,127-rcm-lysozyme-Lysozyme was partially reduced and carboxymethylated by an adaptation of the method of Acharya and Taniuchi (18).3 Briefly, 0.15 mM lysozyme was treated with 2 mM DTT in 0.1 M Tris acetate, pH 7.8. After 40 min a t room temperature, 0.5 M K+-iodoacetate was added to a concentration of 12 mM and allowed to react for 30 min in the dark. The protein was then desalted, also in the dark, by elution with 0.1 N acetic acid through a column of Sephadex G-15. After dialysis against 25 mM Napi, pH 6.4, the product was separated from unreacted lysozyme by cation-exchange HPLC on a PolyCAT ATM column (0.46 X 20 cm, 5 pm particle size; The Nest Group) eluted with a linear gradient of 0-0.5 M NaCl in the phosphate buffer. As detected by absorbance a t 280 nm, the product emerged at approximately 0.26 M NaCl and unreacted lysozyme a t 0.35 M NaC1. After concentration and desalting in a Centricon 10 microconcentrator (Amicon Corp.), approximately 1.5 mg of the 6,127-rcm derivative was recovered from 40 mg of lysozyme. The identity of the product was confirmed by amino acid and sequence analyses (see below and under "Results").
Iodoacetate Trapping of Free Sulfhydryl Groups-To detect reduced species of lysozyme produced by various concentrations of DTT under conjugation assay conditions, lysozyme a t 0.03 mM was incubated for 1 h at 37 "C in 50 mM Na+-Hepes, pH 7.2, with 0, 2, 4, 8, 12, or 16 mM DTT. The pH of each reaction was then adjusted to 7.8 by the addition of 1 M Tris, pH 9.0, followed immediately by enough 1.5 M K+-iodoacetate in 25 mM Tris acetate, pH 7.8, to give a final iodoacetate concentration of 0.5 M. After 15 min at room temperature in the dark, the reaction was diluted 10-fold into cold 25 mM Napi, pH 5.0, and analyzed immediately with the PolyCAT A cation-exchange HPLC system described above. The percentage of protein emerging This reference reports formation of a two-disulfide form of lysozyme. However, M. Denton and H. A. Scheraga have found (personal communication), and we have confirmed, that a three-disulfide lysozyme derivative is the major product. as a rcm/three-disulfide lysozyme derivative was calculated from the relative peak areas.
Analytical Methods-Protein concentrations were determined spectrophotometrically with extinction coefficients a t 280 nm of 0.16 ml mg-' cm" for ubiquitin (19) and 2.64 ml mg" cm" for lysozyme (20). The lysozyme extinction coefficient was found to apply as well to 6,127-rcm-lysozyme. This was established by determination of 6,127-rcm-lysozyme stock solution concentrations with the trinitrobenzene sulfonate assay (21) using lysozyme as a standard.
Amino acid compositions were determined with the o-phthalaldehyde precolumn derivatization method (22) as described (9). Protein sequencing was done at the UCLA Protein Microsequencing Laboratory with an Applied Biosystems model 470A gas-phase sequenator equipped with a 120A phenylthiohydantoin analyzer for on-line HPLC detection. Prior to sequencing, some samples were fully reduced and S-pyridylethylated with 4-vinylpyridine (Aldrich) (23).
Circular dichroism spectra were acquired with a Jasco 5-600 spectropolarimeter. The instrument was calibrated with (+)-10-camphorsulfonic acid (24), and protein solutions of 0.3-0.5 mg/ml were used in a 0.02-cm path length quartz cell (Hellma Cells, Inc.). Typically, a 10 nm/min scan rate with a 4-s time constant was used, and 12 scans were averaged for each spectrum.
Gel filtration analyses were done with a TosoHaas G2000-SWXL column (0.78 X 30 cm) eluted a t 0.4 ml/min with 0.2 M Napj, pH 7.0. An HPLC system equipped with Waters 510 pumps and a Shimadzu SPD-6AV detector was employed.

Conjugation of Hen (Chicken) Egg White Lysozyme with
Ubiquitin Variants in Reticulocyte Lysate-Previous studies have shown that ubiquitin-lysozyme conjugates, containing multiple ubiquitin moieties, are formed prior to the degradation of lysozyme in reticulocyte lysates (16,25). At least a portion of the ubiquitin moieties are in the form of ubiquitinubiquitin linkages, given that the number of ubiquitins can far exceed the total of 6 lysine residues in lysozyme (26, 27). Furthermore, the number of ubiquitin moieties in these conjugates was reduced when ubiquitin was replaced with an Nmethylated ubiquitin derivative (27). We previously have described a set of ubiquitin variants that can be used to test for the presence of specific ubiquitin-ubiquitin linkages in protein conjugates (13). Fig. 1 depicts the alterations that were introduced into ubiquitin to generate the variants used in this study. The proteins all were formed by chemical modification of the site-specific mutant protein, Ub-C48, in which Lys4' of wild-type ubiquitin has been replaced by a cysteine. Neither MeUb-C48 nor MeUb-MeAEtC48 can form ubiquitin-ubiquitin linkages due to the lack of free amino groups. Ub-C48 has the potential of forming wild-type ubiquitin-ubiquitin linkages except at residue position 48, whereas linkage to MeUb-AEtC48 is restricted to the S-aminoethylcysteine at position 48. Native or variant ubiquitin was added to a ubiquitin-depleted reticulocyte lysate to test for the elaboration of specific multiubiquitin chain structures onto lysozyme, and the results are described below. Fig. 2A shows the autoradiograph of an SDS-polyacrylamide gel of the ubiquitin-lysozyme conjugates that were formed between '251-lysozyme and ubiquitin or one of the ubiquitin variants. With native ubiquitin, as had been shown previously by others (16), distinct ubiquitin-lysozyme conjugates can be resolved according to their molecular sizes (Fig.  2 A , left panel, lane 2 ) . When the ubiquitin-depleted reticulocyte lysate was supplemented with either MeUb-C48 or MeUb-MeAEtC48, neither of which is capable of ubiquitinubiquitin linkage, four conjugates were detected (Fig. 2 A , left  panel, lanes 3 and 5 ) . These species migrated with apparent molecular masses of 23, 29, 38, and 41 kDa, consistent with conjugates having one to four ubiquitin moieties, respectively. This result suggested that at least 4 lysines on lysozyme can be linked with ubiquitin. With native ubiquitin, conjugates of greater molecular weights than were found with the N-methylated variants indicate formation of multiubiquitin chains. That the ubiquitin-ubiquitin linkages in the lysozyme conjugates are formed exclusively through an isopeptide bond between the C-terminal carboxyl group of one ubiquitin and the c-amino group of Lys4' in another ubiquitin is suggested by two lines of evidence. First, the conjugates obtained with either Ub-C48 or MeUb-C48 were indistinguishable on SDSpolyacrylamide gels (data not shown), indicating that ubiquitin lysines other than Lys4' did not contribute to ubiquitinubiquitin linkages. Second, conjugates with the MeUb-AEtC48 variant included a number of high molecular weight species found with wild-type ubiquitin but not the other derivatives.
Ubiquitin-dependent degradation of lysozyme is shown in Fig. 2B. Only MeUb-AEtC48 was comparable (within 10%) to native ubiquitin in its ability to stimulate lysozyme degradation in the ubiquitin-depleted reticulocyte lysate, whereas the other three ubiquitin variants stimulated degradation by no more than 30% (data not shown). This result is in agreement with previous studies (10) that suggested that a specific multiubiquitin chain is responsible for the proteolytic targeting of a ubiquitin-protein conjugate.
The formation of ubiquitin-lysozyme conjugates exhibited a strong dependence on the presence of the reducing reagent, DTT. Conjugate formation decreased greatly when 0.01 rather than 2 mM D T T was used (compare left and right panels, Fig.  2 A ) ; a similar D T T dependence was observed when conjugation assays employed unlabeled lysozyme and either lZ5I-or [le~cyl-~H]ubiquitin (data not shown). Likewise, D T T was essential for ubiquitin-dependent proteolysis of '251-lysozyme (Fig. 2B). Although ubiquitin conjugation enzymes and the protease component in the reticulocyte lysate possess critical cysteines that must be preserved for their activity, two lines of evidence suggested that the primary effect of the reducing agent in these experiments is to convert lysozyme from an inactive to an active substrate. First, DTT-depleted and 2 mM DTT-supplemented reticulocyte extracts were equally capable of ubiquitin conjugation to several ribonuclease A derivatives as well as numerous endogenous proteins in the lysate (9). Similarly, ubiquitin conjugation and ubiquitin-mediated degradation of calmodulin were equally efficient with either 0.01 or 2 mM DTT (not shown). Second, increasing concentrations of DTT lead to concomitant increases in the formation of ubiquitin-lysozyme conjugates. Experiments that demonstrate this are described below. Specific Reduction of the C y~~-C y s '~~ Disulfide Bond in Lysozyme-To examine the effect of DTT on lysozyme, we first treated the protein with 2 mM DTT, a concentration found to be effective for the ubiquitin-dependent degradation of lysozyme, under conditions of pH and ionic strength which mimicked those of conjugation and degradation assays. The reduction of one or more of the four disulfide bonds in lysozyme (Fig. 3A) was assessed by the use of iodoacetic acid to trap any free cysteines generated and by cation-exchange chromatography to resolve rcm species (28) from native lysozyme. Fig. 3B shows the elution profile of lysozyme before and after treatments with DTT and iodoacetic acid. With 2 mM DTT, a minor peak, corresponding to about 1.2% of the total lysozyme, was found to elute at a position consistent with the incorporation of carboxymethyl groups into lysozyme. This species, which doubled when 4 mM DTT was used (Fig. 3B, inset (a)), coeluted with the partially reduced and carboxymethylated lysozyme prepared as described under "Experimental Procedures" (not shown). Amino acid analysis of this protein revealed an average of 1.8 carboxymethylcysteine residues per lysozyme molecule, consistent with reduction of one of the four disulfides. That one disulfide was reduced selectively was apparent from N-terminal protein sequencing, which yielded a carboxymethylcysteine at residue position 6. However, because the yield of this compound was not quantitative and cysteine or cystine, even if present at position 6, would have escaped detection, the possibility remained that other three-disulfide lysozyme species had formed as well. To test this, the isolated rcm/three-disulfide lysozyme was reduced completely and pyridylethylated (see "Experimental Procedures"). This derivative and a completely reduced and pyridylethylated lysozyme control were then sequenced through eight Edman degradation cycles. Whereas S-pyridylethylcysteine was detected at position 6 of the control lysozyme derivative, carboxymethylcysteine was found exclusively, with no trace of the pyridylethyl compound, at position 6 of the three-disulfide lysozyme sample. Thus, the small amount of reduced lysozyme formed under conjugation assay conditions is cleaved specifically at the C y~~-C y s '~~ disulfide, and this partially reduced protein could be trapped to yield 6,127-rcm-lysozyme.
The iodoacetate trapping/HPLC method was used to quantify the fraction of three-disulfide lysozyme formed at different concentrations of DTT, and the results are shown in Fig.  4. As expected, the reduced lysozyme increased linearly with up to 8 mM DTT. Beyond 10 mM DTT, however, the yield of the three-disulfide species began to plateau (not shown), a result we attribute to increasing aggregation of proteins in the reaction. If the three-disulfide derivative, but not native lysozyme, is the true ubiquitination substrate, then conjugate yields should show a similar dependence upon DTT concentration. Conjugation assays with Iz5I-ubiquitin and lysozyme were performed with various levels of DTT, and the conjugates were quantified by excision and y-counting of the appropriate band from an SDS-polyacrylamide gel. The result, shown for tetraubiquitinated lysozyme, is that conjugation parallels generation of the three-disulfide form of lysozyme (Fig. 4) and supports our conclusion that cleavage of the CysG-CyslZ7 disulfide is the obligatory step for conversion of lysozyme into a ubiquitination substrate.
When 6,127-rcm-lysozyme was tested as a substrate instead of native lysozyme, we found that both its ubiquitin conjugation and its degradation in reticulocyte lysate did not require DTT (Fig. 5): Moreover, the rate and extent of both ubiquitin conjugation and degradation of the modified lysozyme were considerably higher than those of its native counterpart in the presence of 2 mM DTT (Fig. 6). Thus, prior reduction of the C y~~-C y s '~~ disulfide bond not only circumvented the requirement for a reducing agent in the assay, but the modified ' It should be noted that ubiquitin conjugation to 6,127-rcm-lysozyme decreased with time in the lysate, even without added ATP, as determined from lowered conjugate yields following preincubation of the substrate in the lysate prior to ubiquitin and ATP addition. This inactivation was much faster when 2 mM DTT was included. Whether this effect is due to further reduction of disulfide bonds in the 6,127rcm-lysozyme, nonspecific proteolysis, or other causes remains to be determined. lysozyme also behaved as a kinetically competent intermediate. These results suggest that the rate-limiting step in the degradation of lysozyme in reticulocyte lysate is the reduction of the C y~~-C y s '~~ disulfide bond. To validate this conclusion, it was necessary to demonstrate that the reduction and the carboxymethylation of Cys' and CYS''~ did not lead to an alternative pathway for ubiquitination of the modified lysozyme. Experimental support for this point is described below.
Previous studies had shown that an early event in the recognition of lysozyme by the ubiquitin pathway is the specific binding of the ubiquitin-protein ligase, E3, to the Nterminal lysine on lysozyme (6). The importance of this pathway for lysozyme ubiquitination in our experiments was apparent from comparative studies with ring-necked pheasant lysozyme. This lysozyme, which has a 93% sequence identity with the chicken protein, has Gly (rather than Lys) as its first amino acid (30). As expected for E3-dependent ubiquitination and a protein with a nonpermissive N-terminus (4), virtually no conjugation to the pheasant lysozyme was observed in the reticulocyte lysate, with or without DTT (data not shown). The conclusion, that lysozyme ubiquitination in these reactions was predominantly E3-dependent, was extended to was performed in the reticulocyte lysate containing 2 mM DTT and 50 PM of either native or modified lysozyme radioiodinated to the same specific radioactivity (1.5 X 10" cpmlpg). The reactions were terminated at the times indicated (min) by the addition of SDS sample buffer, separated by electrophoresis on a 12% SDS-polyacrylamide gel, and autoradiographed. Arrows on the right indicate the migration positions of molecular weight markers as in Fig. 2. HEWL, I 25 I-lysozyme.
6,127-rcm-lysozyme by the use of a Lys-Ala dipeptide to inhibit E3 binding to the substrate. Fig. 7 shows that this dipeptide inhibits both the conjugation and the degradation of 6,127-rcm-lysozyme, as had been reported previously for native lysozyme (6). Thus, reduction of the Cy~~-Cys'~' disulfide bond and carboxymethylation of the cysteines did not alter the initial E3 recognition step in the reaction pathway. We note that, without DTT, even a 20-fold excess of native lysozyme did not inhibit ubiquitin conjugation to the 6,127rcm derivative (data not shown). The results presented here, therefore, strongly suggest that the modified lysozyme mimics an intermediate that is generated during the degradation of lysozyme in reticulocyte lysates. The 6,127-rcm-lysozyme Derivative Has a Native-like Structure-The conformation of 6,127-rcm-lysozyme was evaluated by comparison of its circular dichroism spectrum with that of native lysozyme. As can be seen in Fig. 8, the two proteins appear nearly identical with respect to their overall secondary structures. Coelution upon gel filtration on a G2000-SWXL HPLC column (0.78 x 30 cm) showed both proteins to have similarly compact structures (data not shown). In these respects, lysozyme offers a very different system than was found with ribonuclease A, where all modified forms that were ubiquitination substrates were virtually devoid of any secondary structure characteristic of the native protein (9):

DISCUSSION
The faithful selection of damaged or aberrant proteins, but not their normal counterparts, is an essential feature in ubiquitin-mediated proteolysis. The basis of this differentiation can, in principle, be understood by determining those structural changes that are responsible for the conversion of a normally stable protein into an active substrate for degradation in this pathway. In the present study, we have shown that the specific reduction of the Cy~~-Cys'~' disulfide bond ' R. L. Dunten and R. E. Cohen, unpublished observations. Degradation assays were as in Fig. 5B and contained 2 mM DTT. The solid lines denote the presence of ATP and MeUb-AEtC48 with (0) or without (0) 2 mM Lys-Ala in the reaction mixture. The dashed line denotes assays where MeUb-AEtC48 and Lys-Ala were both omitted. The inclusion of 2 mM Lys-Ala did not affect the basal degradative rate (data not shown).
in lysozyme converts this protein from an inactive to an active substrate for ubiquitination and its subsequent degradation in a reticulocyte lysate.
The use of a crude reticulocyte lysate to study substrate selection was convenient in that it allowed for the evaluation of proteins with respect to both ubiquitination and degradation in the same reaction mixture. However, a major concern regarding the use of such a crude assay system arises from the inevitable presence of multiple ubiquitin conjugation enzymes, many of which are believed to conjugate ubiquitin in a manner that does not lead to the degradation of the acceptor proteins (2,31). The possibility that a degradation substrate protein can be ubiquitinated at multiple sites by different conjugation enzymes and that not all ubiquitination products are necessarily relevant to proteolysis must be considered. Because of these concerns, we have first established in this study the relevance of the ubiquitin-lysozyme conjugates that FIG. 8. Circular dichroism spectra of lysozyme and 6,127rcm-lysozyme. Proteins were 0.022 mM in 10 mM potassium phosphate, pH 7.2, and a 0.02-cm path length cell was used. Spectra, plotted as the differential molar circular dichroic extinction coefficient (Af) uersus wavelength, are shown for native lysozyme (-) and the 6,127-rcm derivative ( + . . .).
we obtained in the crude reticulocyte lysate. We have demonstrated here that the proteolytic targeting of lysozyme by ubiquitin conjugation, like the previously studied &-galactosidase proteins (lo), requires the formation of a multiubiquitin chain. Our studies indicated that, although a t least four lysine sites on lysozyme can be ubiquitinated, the attachment of ubiquitin to any of these sites is dependent on the binding of the ubiquitin-protein ligase, E3. Because E3-dependent ubiquitination of proteins thus far is characteristic of ubiquitinmediated proteolysis (2,31), our results suggest that multiple lysine sites on lysozyme may be used in the degradation of lysozyme. The role of these conjugation sites in serving as initiation points for multiubiquitin chain synthesis and proteolytic targeting is being evaluated in a separate study. It is important to point out here, however that the reduction and carboxymethylation of C y~~-C y s '~~ did not alter the ubiquitination sites in lysozyme. 6 The observation that ubiquitin conjugation to lysozyme and the ubiquitin-dependent degradation of this substrate protein were both greatly decreased in a DTT-depleted reticulocyte lysate suggested that the reducing agent may act to convert lysozyme from a largely, if not entirely, inactive substrate (see discussion below) to an active substrate. The alternative explanation, that the enzyme components had been rendered inactive due to the absence of DTT, can be ruled out since ubiquitin conjugation to several proteins, including ribonuclease A derivatives and calmodulin, was not affected. Because four disulfide bonds are present in native lysozyme, a likely possibility is that the effect of D T T is due to reduction of one or more of these disulfide bonds. The results obtained in this study are all consistent with this hypothesis. That the reduction of the C y~~-C y s *~~ disulfide is the relevant intermediate in the degradation of lysozyme is suggested by our finding that this disulfide bond is reduced preferentially under the conditions normally employed for degradation assays in reticulocyte lysate. Further support is provided by the demonstration that the 6,127-rcm-lysozyme behaves as a kinetically competent intermediate for both conjugation and degradation. A more rigorous proof of this hypothesis will require direct demonstration of the specific reduction of this disulfide bond in the ubiquitin-lysozyme conjugates. Nevertheless, the present results do indicate that the reduction of this most susceptible disulfide bond in lysozyme is sufficient to convert this protein to a highly active form for ubiquitin-mediated proteolysis.
By kinetic criteria, the 6,127-rcm-lysozyme is a far superior