Kinetic Analysis of the Conjugation of Ubiquitin to Picornavirus 3C Proteases Catalyzed by the Mammalian Ubiquitin-protein Ligase E3 (cid:1) *

The 3C proteases of the encephalomyocarditis virus and the hepatitis A virus are both type III substrates for the mammalian ubiquitin-protein ligase E3 (cid:1) . The conjugation of ubiquitin to these proteins requires internal ten-amino acid-long protein destruction signal sequences. To evaluate how these destruction signals modulate interactions that must occur between E3 (cid:1) and the 3C proteases, we have kinetically analyzed the formation of ubiquitin-3C protease conjugates in a reconstituted system of purified E1, HsUbc2b/E2 14Kb , and human E3 (cid:1) . Our measurements show that the encephalomyocarditis virus 3C protease is ubiquitinated in this system with K m (cid:2) 42 (cid:3) 11 (cid:4) M and V max (cid:2) 0.051 (cid:3) 0.01 pmol/min whereas the parameters for the ubiquitination of the hepatitis A virus 3C protease are K m (cid:2) 20 (cid:3) 5 (cid:4) M and V max (cid:2) 0.018 (cid:3) 0.003 pmol/min. Mutations in the destruction signal sequences resulted in changes in the rate at which E3 (cid:1)

The selection of proteins for destruction by the ubiquitin 26 S/proteasome pathway depends upon specific interactions that occur between the targeted substrates and enzymes involved in the formation of the ubiquitin-target protein conjugates. A hierarchical family of pathways, each composed of at least three enzymes, accomplishes the attachment of ubiquitin to proteins destined to be degraded (1)(2)(3). Common to all of these pathways is the ubiquitin-activating enzyme, E1, 1 which recruits free ubiquitin through the ATP-dependent formation of a thiolester bond between a cysteine in the E1 and the C-terminal glycine of the ubiquitin molecule. This ubiquitin is then transferred to one of several members of the E2 family of proteins that are referred to as ubiquitin carrier proteins or ubiquitin-conjugating enzymes. Finally, the ubiquitin is transferred from the E2 to the target substrate protein through the action of an ubiquitin-protein ligase, or E3. Although each E2 protein appears to function with several specific ubiquitinprotein ligases, each E3 can specifically interact with only a limited number of substrate proteins. Regardless of the E3 involved in the ubiquitination process, following the conjugation of the first ubiquitin molecule to a primary amine on the substrate protein, the E3, or the E3 plus E2 proteins, can catalyze additional conjugating reactions that result in the synthesis of a chain of ubiquitin molecules attached to the substrate (4).
Important unanswered questions remain as to precisely how the E3 ubiquitin-protein ligases recognize and interact with their substrate proteins. It appears that proteins degraded by the ubiquitin/26 S proteasome system contain structural features, often short primary sequence elements (2,5), that act as protein destruction signals, and presumably it is these structural features that serve as sites for interaction with specific E3 proteins. Very few precisely mapped protein destruction signal structures have been matched with their cognate ubiquitin-protein ligase, however (2). Among the most well studied E3 proteins are mammalian E3␣, which functions in conjunction with the ubiquitin carrier protein HsUbc2 (E2 14K ; see Refs. 6 and 7), and the yeast homologue of E3␣, Ubr1p, which requires the presence of yeast Ubc2p/Rad6 ubiquitin carrier protein (8,9). E3␣ and Ubr1p were first shown to recognize proteins with N-terminal basic (type I) or bulky hydrophobic (type II) amino acids as substrates (10 -16). Based on the affinity resin binding behavior of E3␣ (13), measurements of the degradation rates of artificial substrates in reticulocyte lysate and in intact yeast cells (7,12), and in vitro and in vivo dipeptide competition studies (11,13,14), it was proposed that these enzymes contain both type I and type II N-terminal amino acid binding sites (13,15,16). These binding sites were assumed to provide the means by which substrate proteins are recognized by E3␣ and Ubr1p. In recent years it has been discovered that E3␣ and Ubr1p can catalyze the ubiquitination of proteins lacking destabilizing N-terminal amino acids (type III substrates). The short-lived yeast proteins Gpap and Cup9p, neither of which contains a destabilizing N-terminal amino acid, have been reported to be substrates for Ubr1p (17)(18)(19). Ribonuclease S, the subtilisn-derived fragment of ribonuclease A, has a stabilizing serine N terminus, but it is known to be a substrate for mammalian E3␣-dependent ubiquitin-protein conjugate synthesis (20). This indicates these E3 enzymes can recognize substrate proteins through associations with other types of structural elements.
E3␣ has recently been found to catalyze the conjugation of ubiquitin to two additional proteins that, based upon their N-terminal amino acids, would not be predicted to be N-end rule substrates for degradation. The 3C proteases produced by the encephalomyocarditis virus (EMCV) and the hepatitis A virus (HAV), both members of the picornavirus family, have been shown to serve as substrates for E3␣-dependent ubiquitination (21)(22)(23). 2 The ten-amino acid sequence 34   , located in what is probably a strand-turn-strand structure, has been discovered to function as a protein destruction signal in the EMCV 3C protease (22). The HAV 3C protease contains the sequence 32 LGVKDDWLLV 41 in a location homologous to that of the EMCV protein destruction signal sequence, and this sequence has been shown to be required for the ubiquitination and degradation of the HAV 3C protein (21,23). 3 The identification of two substrate proteins recognized by E3␣, both of which contain precisely mapped, internal sequences known to be required for E3␣-dependent ubiquitin conjugation, provides excellent model systems for detailed studies of the interactions that take place between E3␣ and substrate proteins lacking a destabilizing N-terminal amino acid.
The recent development of an affinity chromatography purification method, based upon the specific binding of mammalian E3␣ to HsUbc2b, 4 has made it possible to obtain sufficient quantities of pure E3␣ to allow biochemically defined kinetic studies of the E3␣-catalyzed conjugation of ubiquitin to different types of substrate proteins (20). We have used a reconstituted system of purified E1, HsUbc2b, and affinity-purified human E3␣ to evaluate the kinetics of the conjugation of ubiquitin to the EMCV and HAV 3C proteases. We have determined the K m and V max values for the E3␣-dependent conjugation of ubiquitin to the wild type 3C proteases and to 3C protease proteins containing mutations in their defined, internally located protein destruction signal regions. The kinetics with which EMCV 3C proteases containing selected lysine to arginine substitutions are ubiquitinated were also evaluated. Our results indicate that differences in the ability of the 3C proteases to serve as substrates for E3␣ are most likely the result of differences in the k cat values with which the catalysis of ubiquitin conjugation occurs. We also obtained data demonstrating that the interaction of E3␣ with the 3C protease proteins involves the same site, or sites, with which it associates with both basic and hydrophobic destabilizing N-terminal amino acids. Our findings explicate the ability of E3␣ to target non-N-end rule substrates for degradation and reveal that the mechanism by which the ubiquitin-protein ligase E3␣ selects substrate proteins for ubiquitination is considerably more complicated than suggested by earlier studies. The construction of the expression plasmids pETE3BЈCDЈ*,  pETHAV3C, and pETP3C have already been described (21-23).  pETE3C A38ϩ , which contains the sequences coding for the EMCV 3C  protease with an alanine inserted between amino acid positions 38 and  39, pETE3C L34A , pETE3C R39D , pETE3C K10,14R , pETE3C K74,77R , and pETE3C K98,101R were prepared using polymerase chain reaction-based oligonucleotide-directed mutagenesis. pE3C (23) was employed as a template, and DNA insert fragments containing the mutated EMCV 3C protease coding sequences were synthesized using the appropriate mutagenic primer (CTTGCCTCCTTGTGAGAGGCGCCCGCACCTTGGT-AGTTAATAG for pETE3C A38ϩ , GAGGCCGCACCGCGGTAGTAAATA-GACACATG for pETE3C L34A , and CTTCTTGTGAGAGGCGACACCT-TGGTAGTAAATAG for pETE3C R39D ). The inserts were ligated into pET3d at the NcoI and BamHI sites (24). Likewise, the expression plasmids pETHAV3C A37ϩ , which contains the sequences coding for the HAV 3C protease with an alanine inserted between amino acid positions 37 and 38, and pETHAV3C D37R were prepared using polymerase chain reaction. pHAV3C (21) was employed as a template, and DNA insert fragments containing the mutated HAV 3C protease coding sequences were synthesized using the appropriate mutagenic primer (CT-TGGGAGTGAAAGATGATGGCTGGCTGCTTGTGCCTTC for pETHA-V3C A37ϩ , and CACAAGCAGCCAACGATCTTTCACTCCCAAG for pET-E3C D37R ). The inserts were ligated into pET3d. pETP3C N33R , which contains the sequences coding for a mutated poliovirus 3C protease, was constructed using pETP3C as a template. A DNA insert fragment containing the mutated poliovirus 3C protease coding sequence was synthesized using the mutagenic primer GGAGTCCACGACCGCGTG-GCTATTTTACCAACC. The insert was ligated into pET3d at the NcoI and BamHI sites.

Construction of Plasmids Containing 3C Protease Coding Sequences-
Purification of E1, HsUbc2b, E3␣, and EMCV and HAV 3C Protease Proteins-Human erythrocyte fraction II was prepared using procedures described previously (25). Human ubiquitin-activating enzyme E1 was purified from this material using a ubiquitin affinity column and fast protein liquid chromatography methods described previously (26). Human recombinant HsUbc2b (27) was expressed in Escherichia coli and purified using methods reported previously (20). Some of this protein was employed in the preparation of an affinity column, which was then used to purify ubiquitin-protein ligase E3␣ from the fraction II preparation as reported recently (20).
The wild type EMCV 3C and mutated (3C(ϩA38), 3C(L34A), 3C(R39D), 3C(K10,14R), 3C(K74,77R), and 3C(K98,101R)) protease proteins were expressed in E. coli from pETE3BЈCDЈ*, pETE3C ϩA38 , pETE3C L34A , and pETE3C R39D and were purified from refolded inclusion body material using procedures reported previously (22). Wild type HAV 3C protease was expressed in E. coli from pETHAV3C and purified as described previously (21). The mutated HAV 3C(ϩA37) and 3C(D37R) proteins were expressed from pETHAV3C A37ϩ and pETE3C D37R , but the presence of the mutations necessitated an altered purification scheme. Cleared lysates from induced cells were passed through a Q-Sepharose column (Amersham Pharmacia Biotech) equilibrated in TDE buffer (50 mM Tris-HCl, pH 7.6, 1 mM dithiothreitol, and 0.1 mM EDTA) containing 0.1 mM phenylmethylsulfonyl fluoride. The bound 3C protease proteins were eluted with a gradient of 300 to 650 mM NaCl in TDE buffer. The column fractions containing the 3C protease proteins were fractionated further and concentrated by precipitation with 30 to 50% (NH 4 ) 2 SO 4 . The precipitates were resuspended in 10 mM KH 2 PO 4 -K 2 HPO 4 , pH 6.9, and applied to a column of Bio-Gel HTP hydroxyapatite (Bio-Rad) equilibrated in the same buffer. Bound material was eluted with a step gradient of 100 to 300 mM KH 2 PO 4 -K 2 HPO 4 , pH 7.2. The eluted proteins were dialyzed against TDE buffer containing 10% glycerol. Stock solutions of the 3C protease preparations were prepared to be Յ 2 mg/ml. We have observed that at least some of the 3C proteins form aggregates, detectable by size exclusion chromatography, during long term storage at greater concentrations than this.
All of the purified 3C protease proteins, with the exception of the wild type EMCV 3C protease, have an N-terminal methionine instead of the naturally occurring glycine or serine residues found in the mature proteins. The non-mutated EMCV 3C protease expressed in E. coli from pETE3B'CD'* undergoes self-processing to produce a protein with a glycine residue at the N terminus (22). For convenience, both nonmutated proteins are referred to here as wild type proteins.
Measurements of the Rates of 125 I-Ubiquitin-3C Protein Conjugate Formation and K m and V max Determinations-Bovine ubiquitin (Sigma) was purified further to apparent homogeneity (28) and radioiodinated by the Chloramine T procedure (14). Some of this material was subjected to reductive methylation (29). The initial rates of 125 I-ubiquitin conjugation were measured using an adaptation of methods employed previously (20). The reaction mixtures typically contained, in a final volume of 25 l, 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 10 mM creatine phosphate, 2 mM ATP, 1 mM dithiothreitol, 1 international unit/ml creatine phosphokinase (Sigma), 1 international unit/ml high protein liquid chromatography-purified yeast inorganic pyrophosphate (Sigma), 50 nM purified E1, 500 nM purified HsUbc2b, 0.2 g of affinitypurified E3␣ preparation, 0 to 20 M exogenous protein substrates, and 4 M 125 I-ubiquitin or 125 I-methylated ubiquitin (about 12,000 cpm/ pmol). The mixtures were incubated for 15 min at 37°and then boiled for 4 min in the presence of 25 l of added sample buffer. This incubation time was selected to yield a linear initial rate of monoubiquitina-tion. The samples were analyzed by 12% SDS-PAGE and autoradiography. The amounts of monoubiquitin-substrate protein conjugates formed were determined by cutting slices from the dried gels and subjecting them to ␥-counting (30). Control experiments confirmed that the initial rate was linear with [E3␣] 0 and independent of [E1] 0 or [HsUbc2b] 0 (20).
Reaction rate data sets were generated by measuring the initial rates of monoubiquitinated conjugates produced in several simultaneously incubated reaction mixtures containing varying concentrations of 3C protease substrate. Two to four data sets were generated for each substrate protein using the same preparations of E1, HsUbc2b, and E3␣. The rate versus substrate concentration data sets were simultaneously fit for each substrate to the K m value using a non-linear least squares regression analysis program (Sigma Plot 5.0). The V max values for each substrate were calculated by averaging the values derived from the fits for each data set.

Evaluation of the Susceptibility of Poliovirus 3C Protease Proteins toward Conjugation with Ubiquitin in Reticulocyte Lysate-A coupled
in vitro transcription-translation rabbit reticulocyte system (Promega) was employed to prepare 35 S-labeled poliovirus 3C and 3C(N33R) proteins encoded within pETP3C and pETP3C N33R . The ability of these proteins to serve as substrates for ubiquitination was evaluated by incubating 7 l of transcription-translation reaction mixtures in a final volume of 20 l containing 20 mM HEPES-KOH, pH 7.5, 1 mM dithiothreitol, 0.1 mM methylated ubiquitin, 0.1 mg/ml cycloheximide, and 60% by volume reticulocyte lysate containing an energy-generating system at 30°for 40 min (21)(22)(23). Aliquots of the reaction mixtures were analyzed by 12% SDS-PAGE and fluorography.

Kinetic Characterization of E3␣-catalyzed Conjugation of Ubiquitin to the 3C Proteases and to 3C Proteases with Protein
Destruction Signal Mutations-Under appropriate conditions, kinetic analysis provides a sensitive and accurate means of quantifying enzyme-substrate interactions and the catalytic competence of the resulting Michaelis complex. We employed a biochemically defined, reconstituted N-end rule ubiquitin ligation system, comprised of affinity-purified human E1 and E3␣ and recombinant human HsUbc2b, to quantitatively evaluate the targeting of the EMCV and HAV 3C proteases for ubiquitin attachment and to assess the effect disrupting their respective destruction signal sequences has on their selection as N-end rule pathway substrates.
A comparison of the labeled products generated by the reconstituted system with either ␣-lactalbumin or wild type EMCV 3C protease as the substrate is shown in the autoradiogram in Fig. 1A, lanes 1-3. In the absence of a substrate protein, the otherwise complete assay mixture containing E1, HsUbc2b, and E3␣ catalyzes the synthesis of hyperconjugates to trace protein contaminants in the enzyme preparations, seen at the top of lane 1 (see Fig. 1A and Ref. 20). Monoubiquitin-HsUbc2b conjugates are also generated under these conditions, at the concentration of this enzyme employed to assure E3␣limiting conditions. In reaction mixtures in which ␣-lactalbumin was the substrate most of the products synthesized during the incubation period were high molecular mass polyubiquitin-␣-lactalbumin conjugates, and the characteristic ladder of sequential ubiquitin adducts is apparent (Fig. 1A, lane 2). The conjugation of ubiquitin to the EMCV 3C protease occurred at a markedly lower rate, and the majority of these products consisted of monoubiquitinated 3C protease (Fig. 1A, lane 3). The rate of monoubiquitin-3C protease synthesis was found to be linear for up to 15 min. An increasingly large fraction of the products consisted of polyubiquitinated 3C protease at longer times (data not shown). This suggests the attachment of the first ubiquitin molecule to the 3C protease occurs more slowly than subsequent polyubiquitinated conjugate synthesis. It should be noted that the synthesis of polyubiquitinated ␣-lactalbumin conjugates has been shown to also be linear with respect to time and E3␣ concentration in a reaction system very similar to the one used here (20). Because our goal in this study was to attempt to detect potentially subtle differences in the kinetics with which related substrate proteins are ubiquitinated by the same ubiquitin-protein ligase, we preferred to avoid artifacts that might result from the use of ubiquitin mutants or derivatives that do not support polyubiquitin chain synthesis. In addition, the rate of the first ubiquitin attachment is more likely than subsequent steps to reflect substrate recognition events mediated by E3␣. To confirm that the formation of polyubiquitinated 3C protease was not a major event during the incubation time, reactions were carried out in which 125 I-methylated ubiquitin was used in place of 125 I-ubiquitin. Measurements of the fraction of the total labeled 3C protease-dependent products that migrated in SDS-PAGE gels above the methylated ubiquitin-3C protease (Fig. 1A, lanes 4 and 5) were found to be virtually identical to the fraction of analogous material synthesized in reaction mixtures containing 125 I-ubiquitin. This indicates the amount of monoubiquitinated 3C protease present in the reaction mixtures at the end of 15 min is not significantly affected by reactions that lead to the synthesis of polyubiquitinated 3C protease. Identical results were obtained using the HAV 3C protease as a substrate. The rate of formation of monoubiquitinated 3C protease proteins was therefore taken to be a valid measure of the initial rate at which E3␣ catalyzes the conjugation of the first ubiquitin to these substrates.
For each kinetic analysis, the quantity of 125 I-ubiquitin incorporated into monoubiquitinated conjugates during the 15min incubation period was measured as a function of 3C-protease concentration. An example of the SDS-PAGE gel analysis from a set of reaction mixtures used in measuring the rate of conjugation of ubiquitin to the wild type EMCV 3C protease is shown in Fig. 1B. The range of concentrations of some of the 3C protease proteins employed in these measurements did not produce initial velocities that approach saturation, which has the potential to introduce error into the calculations of kinetic parameters. Higher substrate concentrations would, however, have required the use of stock 3C protease preparations at concentrations we have observed to sometimes result in the formation of aggregates during long term storage. Despite this limitation to the reaction conditions, the reproducibility of the velocity versus substrate concentration measurements supports the reliability of the data. Fig. 1C shows the simultaneous fit of initial velocity versus substrate concentration data from four sets of measurements using the EMCV 3C protease. As with all of the 3C proteins employed in this study for which rate measurements could be made, the dependence of the initial ubiquitin attachment rate upon substrate concentration exhibited hyperbolic kinetics. This is confirmed by the linearity of the data in the double reciprocal plots, demonstrated in Fig.  1D. The random distribution of the plotted residuals versus the theoretical fit of the data (Fig. 1, C and D) are consistent with the Michaelis-Menten analysis used here to determine the kinetic parameters. The K m and V max values for the conjugation of ubiquitin to the substrate proteins were calculated using non-linear least squares fit analysis of the velocity versus substrate concentration data, as described under "Experimental Procedures." Both the wild type 3C proteases and 3C proteases containing mutations in the protein destruction signal sequence were evaluated as E3␣ substrates. The wild type protein destruction signal region sequences and the sequences present in the mutated proteins used in this study are shown in Fig. 2. The L34A and ϩA38 mutations in the EMCV 3C protease were selected, because proteins carrying these changes have been shown previously to serve as poor substrates for the ubiquitin/26 S proteasome system in rabbit reticulocytes (23). The leucine residue occupying the first position of the sequence has been shown to be particularly important for signal function, as has the distance separating the hydrophobic amino acid triplets on either end of the signal sequence. The ϩA37 mutation in the HAV 3C protease was prepared, because we wished to evaluate the kinetics of an HAV 3C protein carrying a mutation we predicted would render the protein a poor E3␣ substrate by increasing the distance between the distal hydrophobic residues in the destruction signal sequence. The HAV 3C(D37R) mutant was prepared to provide a substrate for testing the effects of replacing one of the two negatively charged amino acids in the HAV 3C protease destruction signal with a positively charged amino acid. This mutated protein carries an arginine residue in the same relative position as the Arg-39 residue in the wild type EMCV 3C protease signal sequence. To serve as a complement for this substrate, the EMCV 3C(R39D) protein was also prepared.
Measurements of the initial rates for the E3␣-dependent conjugation to the wild type and mutated EMCV and HAV 3C proteases were used to calculate the K m and V max values for the reactions with each substrate. The values of the kinetic parameters are displayed, for ease of comparison in Fig. 3A (open bars for K m values, and shaded bars for V max values), and a comparison of the V max /K m ratios is presented in Fig. 3B. Because the same concentration of E3␣ was used in all of the reaction mixtures, V max can be assumed to be proportional to k cat , and V max /K m can be employed for comparing catalytic efficiencies with which E3␣ conjugates ubiquitin to the 3C protease substrates. Also for comparison, an SDS-PAGE analysis of the products generated in reactions containing ␣-lactalbumin and the 3C proteases is shown in Fig. 3C. Again, ␣-lactalbumin was rapidly incorporated into primarily large polyubiquitinated conjugates, whereas the ubiquitinated wild type 3C protease products were synthesized more slowly and included mostly monoubiquitinated conjugates. Differences in the quantity of monoubiquitinated 3C protease products synthesized during the incubation period are evident.
The K m and V max values for attachment of the first ubiquitin to the wild type 3C proteases (42 Ϯ 11 M and 0.051 Ϯ 0.01 pmol/min, respectively, for the EMCV 3C proteases and 20 Ϯ 5 M and 0.018 Ϯ 0.003 pmol/min, respectively, for the HAV 3C proteases) are of similar orders of magnitude as those reported for other substrates of mammalian E3␣ (20). Changes in the protein destruction signal sequences of both 3C proteases were found to have measurable effects on the kinetics with which these proteins are ubiquitinated. Both the K m and V max values for the attachment of the first ubiquitin to the EMCV 3C(ϩA38) and EMCV 3C(L34A) proteases were found to be reduced relative to the values calculated for the reactions with the wild type protein (Fig. 3A). An example of the simultaneous fit of the initial velocity versus substrate concentration data for one of the EMCV 3C protease destruction signal mutants, 3C(ϩA38), is shown in Fig. 4. In this case, the reaction does approach saturation at the higher concentrations of substrate. The HAV 3C(ϩA37) protease was such a poor substrate for E3␣ (evident from the results shown in Fig. 3C) that reliable initial rate measurements were not possible. These results are consistent with the earlier characterization of the 3C protease destruction signal (23). The reactions with the HAV 3C(D37R) protease occurred with K m and V max values both severalfold larger than the corresponding wild type substrate parameters (Fig. 3A). Although the data clearly indicate the D37R mutation results in large increases in K m and V max for the ubiquitination of the protease, the high parameter values mean that the substrate concentrations in the reaction mixtures were far from saturating. These values therefore have a relatively large associated uncertainty. These results nevertheless indicate that a basic amino acid in this position positively affects interactions between the destruction signal region and E3␣. The complementary reverse mutation in the EMCV 3C(R39D) protein did not, however, lead to a reduction in the kinetic parameter values. This substrate was ubiquitinated in the reconstituted system with K m and V max values similar to those for the wild type protein (Fig. 3A). As Fig. 3B shows, the V max /K m ratios for the ubiquitination of the wild type and mutated 3C proteases are similar.
Although the differences between the K m and V max values for the wild type and mutated 3C proteases are not large, they likely reflect genuine differences in the interactions between these proteins and E3␣. A previously published demonstration that mutations in the destruction signal region of the EMCV 3C protease can interfere with the E3␣-catalyzed ubiquitination of this protein (23) supports the quantitatively derived results obtained here. It is unlikely the kinetic differences are due, for example, to the variable presence of small amounts of nicked or truncated species in the protein preparations that resulted in the exposure of types I or II substrate N-terminal amino acids. The same preparations of E1, HsUbc2b, E3␣, and energy system enzymes were used for all of the measurement reported here. Two different preparations of the wild type HAV 3C protease were used in the rate measurements performed for this study, and both were ubiquitinated with very similar kinetic parameters. This indicates that if contamination of the protease preparations occurs, it occurs in a reproducible fashion. If the 3C proteases are nicked by proteases during purification, it seems likely that peptide bond cleavage would occur in the same locations and with similar frequencies in both the wild type and mutated protein preparations. Finally, the dramatic reduction in the rate of ubiquitination of the HAV 3C protease caused by the ϩA37 mutation was not accompanied by the obvious synthesis of any other ubiquitin-protein conjugates (Fig. 3C), making it appear unlikely that competition with better E3␣ substrate contaminants is a significant factor.
Attempts were made to measure the rate at which E3␣ catalyzes the conjugation of ubiquitin to the wild type poliovirus 3C protease, but these were unsuccessful. The poliovirus 3C protease has been shown to be a very poor substrate for the ubiquitin/26 S proteasome system (21). The poliovirus 3C protease contains, in a position analogous to that of the EMCV and HAV 3C protease destruction signal sequences, the sequence 28 LGVHDNVAIL 37 (23). Given the results obtained with the HAV 3C(D37R) and EMCV 3C(R39D) proteins, we wondered whether replacing the asparagine residue at position 33 with an arginine would by itself substantially improve the ability of the poliovirus 3C protease to serve as a substrate for the purified E3␣. An arginine residue in this position aligns with both the Arg-39 in the wild type EMCV 3C protease and with the arginine residue substituted into the HAV 3C(D37R) protein (see Fig. 2). Attempts to refold the mutated poliovirus protein from inclusion bodies obtained from expressing E. coli cells were unsuccessful. The poliovirus 3C(N33R) protein was, therefore, instead prepared by in vitro translation, and its ability to serve as a substrate for conjugation with ubiquitin was tested using rabbit reticulocyte cell extracts supplemented with methylated ubiquitin (21)(22)(23). The results of these experiments showed that the N33R substitution had little or no effect on the susceptibility of the poliovirus 3C protein toward ubiquitination (data not shown). Although this result does not prove conclusively that E3␣ fails to interact easily with the poliovirus 3C(N33R) protein, it does suggest that a basic amino acid in this position does not in itself generate a functional protein destruction signal.
Effects of Selected Lysine to Arginine Substitutions on the Kinetics of E3␣-catalyzed Ubiquitination of EMCV 3C Proteases-The EMCV 3C protease contains twelve lysine residues, none of which occur in the destruction signal sequence, and it as been shown that any of these amino acids can serve as the initial ubiquitin-conjugating site (23). It is conceivable that the location of lysine residues in a substrate protein, relative to a protein destruction signal for example, can influence the likelihood with which they are selected by ubiquitin-protein ligases (16,31). To determine whether eliminating some of the potential ubiquitin attachment sites in the EMCV 3C protease changes the kinetics with which E3␣ catalyzes the ubiquitination of this protein, we prepared three EMCV 3C protease proteins that each contain two lysine to arginine substitutions. Based upon sequence alignments with the HAV 3C protease and examinations of the HAV 3C protease three-dimensional structure (32-34), we selected substitution sites likely to exist in separate locations on approximately the same face of the EMCV 3C protein as does the protein destruction signal region. These proteins, EMCV 3C(K10,14R), 3C(K74,77R), and 3C(K98,101R), have been shown previously to retain levels of catalytic activity similar to the wild type 3C protease, indicating these mutations do not induce significant higher order structural alterations (23).
The initial rates at which these mutated proteins are ubiquitinated in the E3␣-dependent ubiquitin-conjugating system were greater than for the wild type protein (Fig. 5), leading to increases in both K m and V max in each case. Although these results provide a clear indication that all three mutations led to a similar effect on the kinetics with which the EMCV 3C protease is ubiquitinated, the calculated K m and V max values contain much larger standard errors, because the K m values are beyond the concentration ranges possible for the 3C protease substrates.
E3␣ Recognizes the 3C Proteases as N-end Rule Substrates-A characteristic feature of E3␣ ligase is its inhibition by dipeptides containing cognate N-end rule residues in the N-terminal position (8,10,11). It was demonstrated recently that this effect follows classic non-competitive inhibition (20), suggesting that the association of substrates with E3␣ involves a site in the ligase distinct from the site that recognizes Nterminal amino acids. We tested whether the EMCV and HAV 3C proteases behave as N-end rule substrates, as defined by the effect dipeptides have on the E3␣-dependent synthesis of ubiquitin-3C protease conjugates. As shown by the data in Fig.  6, both lysylalanine and phenylalanylalanine inhibited the conjugation of ubiquitin to both wild type 3C protease proteins. The ubiquitination of the HAV 3C protease is more strongly inhibited by the dipeptides, especially lysylalanine, than is the EMCV 3C protease. This is probably a reflection of the significant differences in the sequences of these two proteins (32,33), which in turn affects the stability of least one of the 3C protease-E3␣ interactions that must occur during the ubiquitinconjugating mechanism. The inhibitory effects evident when the control dipeptides alanyllysine and alanylphenylalanine were present in the reaction mixtures are because of a reduction in the steady state formation of the activated ubiquitin-E1 thiolester complex (20). DISCUSSION We have used a kinetic approach to examine the interaction of affinity-purified human ubiquitin-protein ligase E3␣ with two naturally occurring N-end rule substrate proteins, the 3C proteases produced by the picornaviruses EMCV and HAV. Both of these proteins behave as type III substrates, because neither has a destabilizing N-terminal basic or bulky hydrophobic amino acid (12,14). Both proteins have instead internal ten-amino acid-long regions that serve as protein destruction signals required for the E3␣-catalyzed conjugation of ubiquitin (21,23). 3 Rate measurements in a reconstituted system under E3␣limiting conditions were used to calculate the kinetic parameters for the E3␣-catalyzed ubiquitination of the wild type EMCV and HAV 3C proteases and 3C proteases containing mutations in the protein destruction signal regions. Conjugation of wild type EMCV 3C protease shows saturation kinetics with a K m of 42 Ϯ 11 M and a V max of 0.051 Ϯ 0.01 pmol/min whereas the HAV 3C protease is ubiquitinated with a K m of 20 Ϯ 5 M and a V max 0.018 Ϯ 0.003 pmol/min. That the conjugation of ubiquitin to these proteins occurs with dissimilar kinetics is not surprising, given that these picornavirus protease orthologs do not share a high degree of sequence homology, and their destruction signal sequences are dissimilar (23,32,33), presumably reflected in the different kinetics of E3␣ ligation. The initial rates with which the 3C proteases are ubiquitinated in the reconstituted system were found to be much lower than for the type I N-end rule substrate ␣-lactalbumin (20), and the major product formed was the monoubiquitinated 3C protease adduct.
Mutations in the destruction signal sequences of both the EMCV and HAV 3C proteases alter the kinetics with which these proteins undergo E3␣-dependent ubiquitination. Replacing the first leucine in the EMCV 3C protease destruction signal sequence with alanine or increasing to five the number of amino acids in the central regions of both the EMCV and the HAV 3C protease signal sequences resulted in proteins ubiquitinated by E3␣ with lower K m and V max values than the wild type proteins. Such mutations in the destruction signal sequence reduce the ability of these proteins to serve as substrates for ubiquitination in reticulocyte lysate (23). An attempt was made to demonstrate that a basic amino acid located in the third position in the central hydrophilic region of the destruction signal sequence enhances signal function in the E3␣ reaction system. Although the replacement of the aspartate at position 37 in the HAV 3C protease with an arginine resulted in an increase in the E3␣-dependent rate of ubiquitination and increases in both the K m and V max for the reaction, the results of experiments with the EMCV 3C(R39D) and poliovirus 3C(N33R) proteases do not support the conclusion that a basic amino acid in this position is a required component of a functional destruction signal. Given that the 3C protease destruction signal is located in a stand-turn-strand motif (23,34), it is conceivable that higher order structure plays a role in the recognition of these proteins by E3␣. It is of interest to note that all of the mutations that resulted in reduced K m and V max values for ubiquitin attachment have also been found to reduce or eliminate the catalytic activities of the EMCV and HAV 3C proteases. 5 Because the destruction signals are located near the catalytic site in both proteins, the effects of the mutations on the reaction kinetics observed here may be due, at least in part, to higher order structural alterations.
The V max /K m ratios for the ubiquitination of the 3C proteases in the reconstituted reaction system are similar, and this provides insights into the mechanistic behavior of E3␣. If it is assumed E3␣ acts upon these substrates in a non-equilibrium, sequential reaction process that can be described by the classic two-step binding and catalysis mechanism, so that V max ϭ k cat [E total ], K m ϭ (k Ϫ1 ϩ k cat )/k 1 , and it is assumed k Ϫ1 Ͻ k cat , then differences in k cat will affect both the V max and the K m proportionally and in the same direction, as observed for the 3C protease substrates. Substrate-dependent variations in k 1 or 5 T. G. Lawson, unpublished results. k Ϫ1 may occur, but these would have to always complement each other in a way that would have a small impact on the K m value, and this is unlikely. The simplest interpretation of the data is that the differences in the rates at which E3␣ catalyzes the initial attachment of ubiquitin to the 3C proteases are the results of mechanistic events that occur subsequent to the initial recognition of the substrate proteins by E3␣. It is possible that enzymatic steps following the initial ubiquitination reaction monitored here can impact the kinetics with which the first reaction occurs, especially if a significant portion of the available E3␣ is involved in the synthesis of polyubiquitinated 3C protease conjugates. Because the synthesis of monoubiquitinated 3C protease conjugates is rate-limiting, and the rate measurements were made prior to the formation of significant amounts of polyubiquitinated products, however, these downstream reaction events should not compromise our interpretation of the data.
The low V max , and therefore low k cat , values that characterize the E3␣-dependent ubiquitination of the 3C proteases in the reconstituted reaction system are not without precedent. The V max for the ubiquitination of ribonuclease S, another type III N-end rule substrate, by E3␣ is similar to those calculated for the 3C proteases (20). These V max values are relatively low compared with the type I and type II substrates ␣-lactalbumin and ␤-lactoglobulin (20). Moreover, the V max /K m ratio for the ubiquitination of ribonuclease S is similar to the ratio found to characterize the reactions with the 3C proteases. ␣-Lactalbumin and ␤-lactoglobulin are ubiquitinated in reactions characterized by V max /K m ratios about 30-and 130-fold greater, respectively (20). This difference in V max /K m values must reflect the effect on E3␣ of the internal position of the recognized destruction signal versus the N terminus of the substrate proteins. The kinetic analysis of the E3␣-dependent ubiquitination of three EMCV 3C protease proteins carrying pairs of lysine to arginine substitutions yielded K m and V max values both greater than for the wild type protein. This result is something of a surprise, because it might be expected that a reduction in the number of available lysine residues would lower the k cat , and thereby the V max , with which ubiquitin conjugation occurs. One possible explanation for these results is that the selection of the initial ubiquitin attachment site is a stochastic process (15), and the active site of E3␣ transiently associates with several available lysine residues until one in a favorable structural context is located. Reducing the availability of less optimal lysine residues could perhaps reduce the time required for E3␣ to locate favored ubiquitin attachment sites. Alternatively, it may be that surface-exposed basic amino acids are specifically involved in the discriminatory interactions that take place between E3␣ and substrate proteins and that E3␣ associates more effectively with arginine residues than with lysines. It can be concluded, at least, that the ten-amino acid protein destruction signal is not the only feature in the EMCV 3C protease capable of modulating E3␣ activity. It is of interest to note that changes in the basic amino acid composition of substrate proteins has also been shown indirectly to affect the activity of yeast Ubr1p, although it is unclear how these changes modulate Ubr1p-substrate interactions (35).
Our analysis of the kinetics of the ubiquitination of the EMCV and HAV 3C proteases indicate that E3␣ interacts with these proteins through a complicated mechanism and that substrate selection is not based solely upon a simple recognition of the substrate to form a ligase-substrate complex. Baboshina et al. (20) have proposed a model by which E3␣ interacts with at least two sites on substrate proteins. Their data suggest that interactions between the N-terminal amino acids of type I or type II N-end rule substrates and a site on E3␣ promote the formation of a catalytically competent conformation of the ligase. This model predicts certain features of type III substrates other than N-terminal amino acids also associate with the N-terminal amino acid binding site and induce, though less effectively, the formation of a catalytically active E3␣. Our results with the 3C proteases are consistent with this model. In the case of the 3C proteins, the destruction signal motifs appear to regulate how rapidly E3␣ can catalyze the conjugation of ubiquitin to the protease proteins. The initial recognition of the 3C protease by E3␣, as defined by k Ϫ1 /k 1 in the two-step reaction model, does not appear to contribute in a major way to the observed differences in the overall rates of initial ubiquitin conjugation. That the 3C proteases must interact with the same site, or sites, on E3␣ that binds basic or bulky N-terminal amino acids is indicated by the fact that both lysylalanine and phenylalanylalanine inhibit the ubiquitination of both the EMCV and HAV 3C proteases. It is conceivable that the hydrophobic and basic amino acids in the destruction signals can mimic destabilizing N-terminal amino acids during the association of the substrates with E3␣. The 3C protease destruction signals do not appear to be as effective as N-terminal basic or bulky hydrophobic amino acids at stimulating E3␣ activity, because the V max values for the E3␣-dependent ubiquitination of types I and II substrates are much higher than for the formation of ubiquitin-3C protease conjugates (20).