Protein interactions within the N-end rule ubiquitin ligation pathway.

Rate studies have been employed as a reporter function to probe protein-protein interactions within a biochemically defined reconstituted N-end rule ubiquitin ligation pathway. The concentration dependence for E1-catalyzed HsUbc2b/E2(14kb) transthiolation is hyperbolic and yields K(m) values of 102 +/- 13 nm and 123 +/- 19 nm for high affinity binding to rabbit and human E1/Uba1 orthologs. Competitive inhibition by the inactive substrate and product analogs HsUbc2bC88A (K(i) = 104 +/- 15 nm) and HsUbc2bC88S-ubiquitin oxyester (K(i) = 169 +/- 17 nm), respectively, indicates that the ubiquitin moiety contributes little to E1 binding. Under conditions of rate-limiting E3alpha-catalyzed conjugation to human alpha-lactalbumin, HsUbc2b-ubiquitin thiolester exhibits a K(i) of 54 +/- 18 nm and is competitively inhibited by the substrate analog HsUbc2bC88S-ubiquitin oxyester (K(i) = 66 +/- 29 nm). In contrast, the ligase product analog HsUbc2bC88A exhibits a K(i) of 440 +/- 55 nm with respect to the wild type HsUbc2b-ubiquitin thiolester, demonstrating that ubiquitin binding contributes to the ability of E3alpha to discriminate between substrate and product E2. A survey of E1 and E2 isoform distribution in selected cell lines demonstrates that Ubc2 isoforms are the predominant intracellular ubiquitin carrier protein. Intracellular levels of E1 and Ubc2 are micromolar and approximately equal based on in vitro quantitation by stoichiometric (125)I-ubiquitin thiolester formation. Comparison of intracellular E1 and Ubc2 pools with the corresponding ubiquitin pools reveals that most of the free ubiquitin in cells is present as thiolesters to the components of the conjugation pathways. The present data represent the first comprehensive analysis of protein interactions within a ubiquitin ligation pathway.

The majority of short-lived cellular proteins are targeted for degradation by the 26 S proteasome in response to assembly of degradation signals on their surface comprising chains of ubiquitin moieties covalently linked through specific lysine residues (1). Target protein specificity for this process is determined in part by a large family of diverse ubiquitin-protein isopeptide ligases (E3) 1 that recognize specific features of the native structure (2)(3)(4), transposable trans-acting amino acid sequences (5-7), or exposed regions of non-native conformation (8). The activated ubiquitin required to drive isopeptide bond formation is donated by specific ubiquitin carrier proteins (E2/ Ubc), also termed ubiquitin-conjugating enzymes, in which the ubiquitin carboxyl terminus is bound as a thiolester to a conserved cysteine (9,10). The apparent specificity of different isopeptide ligases for recognizing a single or limited number of E2 isozymes accounts for the large cohort of related carrier protein isoforms (9 -11). Some E2 moieties may contribute to the substrate specificity of their cognate E3 isozyme because they are able to associate with targets in the absence of ligase (12)(13)(14). In addition, a subset of E2 isozymes catalyzes formation of polyubiquitin degradation signals on model protein substrates in the absence of their cognate E3 (15)(16)(17)(18). The E3independent conjugation reactions catalyzed by these selected isoforms presumably reflect catalytic roles within their respective ligase-dependent substrate targeting mechanisms.
Ubiquitin thiolester formation to the different E2 isozymes occurs by transfer of this moiety from a ternary complex of ubiquitin-activating enzyme (E1) containing two forms of activated ubiquitin polypeptide: a tightly bound ubiquitin adenylate intermediate and a covalent E1-ubiquitin thiolester (19,20). Plants inexplicably contain multiple E1 isozymes (21), whereas other eukaryotes possess a single gene that is transcribed into a 3.5-kb message subject to translation at alternate start sites to yield cytoplasmic and nuclear isoforms of 110 and 117 kDa, respectively (22). The additional amino-terminal 40 residues of the 117-kDa E1 contain a nuclear localization signal and multiple phosphorylation sites that promote sequestering of this isozyme within the nucleus (22). Because each ligase family recognizes a cognate E2 or its orthologs, rates of conjugation through the different pathways and, therefore, their contribution to degradative targeting, depend on efficient loading of these enzymes by the E1 ternary complex (9).
We have shown recently that ubiquitin ligation is amenable to detailed kinetic analysis in biochemically defined in vitro assays (23,24). In the present work, we extend these studies to protein-protein interactions among the components of the Nend rule targeting pathway for ubiquitin conjugation. This pathway requires the E3 isozyme Ubr1 in Saccharomyces cerevisiae (2) and the 19.7-kDa E2 isozyme Rad6/ScUbc2 2 (25). In rabbit reticulocytes the corresponding pathway is catalyzed by E3␣, the mammalian Ubr1 ortholog, in concert with HsUbc2/ E2 14kb (23,26,27), the 17.3-kDa mammalian E2 ortholog of S. cerevisiae Rad6 (9). We have employed rate studies to determine the affinity for binding of HsUbc2b 3 to E1 and of the corresponding HsUbc2b-ubiquitin thiolester to E3␣. In addition, active site mutants of the E2 have been used to estimate affinities for product binding to E1 and E3␣. The latter results have cautionary implications for the unambiguous interpretation of in vivo observations based on overexpression of dominant negative E2 mutants. Finally, activity assays involving 125 I-ubiquitin thiolester formation have been exploited to quantitate intracellular concentrations of E1 and Ubc2 within selected cell lines. Comparison of these binding constants with the concentrations of the components required for N-end ruledependent ligation suggests that the three enzymes of this pathway minimally form dynamic binary complexes within the cell.

MATERIALS AND METHODS
Bovine ubiquitin, creatine phosphokinase, yeast inorganic pyrophosphatase, yeast hexokinase, and human ␣-lactalbumin were purchased from Sigma. The yeast inorganic pyrophosphatase and human ␣-lactalbumin were further purified to apparent homogeneity (23). The ubiquitin was also further purified to homogeneity (15) then radioiodinated by the chloramine-T procedure (28). Restriction endonucleases and other DNA-modifying enzymes were purchased from New England Biolabs or Amersham Biosciences. Carrier-free Na 125 I and [2, H]ATP were purchased from PerkinElmer Life Sciences. Rabbit liver, rabbit reticulocyte, and human erythrocyte E1 enzymes were purified to apparent homogeneity by adapting previously reported affinity chromatography and FPLC methods (29), then quantitated by 125 I-ubiquitin thiolester assay and confirmed by the stoichiometric formation of ubiquitin [ 3 H]adenylate (19,29). Unless otherwise indicated, rabbit reticulocyte E1 was used in all kinetic studies for which the activation step was kinetically isolated, whereas the more abundant but otherwise identical liver enzyme was used for studies in which it served only as a reagent for generating E2 thiolester. Human recombinant HsUbc2b (23) was purified to apparent homogeneity using previously established methods and then quantitated by its stoichiometric formation of 125 Iubiquitin thiolester formation in the presence of E1 (29). Rabbit reticulocyte Fraction II was prepared by phenylhydrazine induction (28). Rabbit reticulocyte and liver E3␣ were purified from Fraction II by E2-ligand affinity chromatography (23). Rabbit reticulocyte E3␣ was used in all kinetic studies for which the ligase step was rate-limiting, whereas the more abundant liver E3␣ was used in coupled assays as a reagent ligase in all other rate studies.
Generation and Purification of Wild Type and Mutant HsUbc2b-The human/rabbit HsUbc2b coding sequence was excised from plasmid pECO10, derived from pKK223-3 (30), and cloned into plasmid pET11d to generate pET-HsUbc2b. Mutation of the active site cysteine 88 of human HsUbc2b to either alanine or serine was achieved by the PCRbased overlap extension protocol of Ho et al. (31) to generate pET-HsUbc2bC88A and pET-HsUbc2bC88S, respectively. The complete HsUbc2b coding regions of these constructs were sequenced using a Sequenase kit (U. S. Biochemical Corp.) to confirm the desired base change(s) and the absence of secondary mutations introduced by the PCR steps.
A single colony of Escherichia coli BL21 (DE3) strain harboring the appropriate plasmid was grown in 10 ml of LB medium containing 100 g/ml ampicillin for 6 h at 37°C, and then 50 l was used to inoculate a 250-ml overnight culture of the same medium. The next morning 10 liters of LB containing 100 g/ml ampicillin were inoculated with 100 ml of the overnight culture and then grown at 37°C to an A 600 of 0.8 -1.0 in a VirTis 20-liter fermenter. Expression was induced by adding isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.4 mM. Following expression for 1.5 h, cells were collected by centrifugation at 4,000 ϫ g for 25 min and then resuspended in 50 mM Tris-HCl (pH 7.5) containing 5 mM EDTA and 5 mM DTT (Buffer A) to ϳ30% (w/v) before lysing by French press. All subsequent steps were carried out at 4°C unless otherwise stated. Recombinant wild type HsUbc2b and HsUbc2bC88S were assayed by their stoichiometric formation of 125 I-ubiquitin thiolester or oxyester, respectively, in brief incubations (29). Recombinant HsUbc2bC88A was assayed by Western blot using an affinity-purified rabbit anti-HsUbc2 antibody and recom-binant HsUbc2b standards (32,33).
The resulting lysate was then centrifuged at 10 5 ϫ g for 1 h after which the high speed supernatant was adjusted to pH 7.5 with 1 M NaOH followed by bulk absorption to 250 ml of DEAE-cellulose (Whatman DE52) equilibrated in Buffer A. The DEAE slurry was stirred gently at 4°C for 1 h to allow complete adsorption of protein and then washed on a Buchner funnel with 3 bed volumes of Buffer A, followed by batch elution with 2 bed volumes of Buffer A containing 0.5 M NaCl. Protein in the eluate was precipitated by adjusting to 80% saturated ammonium sulfate then recovered by centrifugation for 20 min at 10 4 ϫ g. The pellet was suspended in 50 ml of 50 mM Tris-HCl (pH 7.5) containing 1 mM DTT (Buffer B) and then dialyzed overnight against 2 liters of the same. The dialysate was loaded onto a 5 ϫ 10-cm Q-Sepharose Fast Flow (Amersham Biosciences) column equilibrated with Buffer B and fitted to an Amersham Biosciences FPLC system. Protein was eluted with a linear 0 -0.5 M NaCl gradient (1.7 mM/min) in Buffer B at a flow rate of 2 ml/min. Fractions containing HsUbc2b activity (0.18 M elution position) were pooled and adjusted to a final concentration of 1.8 M ammonium sulfate and then applied to a 3 ϫ 25-cm phenyl-Sepharose Fast Flow (Amersham Biosciences) column equilibrated with 1.8 M ammonium sulfate in Buffer B. Protein was eluted with a negative linear gradient of 1.8 -0 M ammonium sulfate (7.2 mM/min) at a flow rate of 2 ml/min. Wild type and mutant HsUbc2b enzymes eluted at ϳ0.5 M ammonium sulfate. Aliquots from the phenyl-Sepharose column containing HsUbc2b were pooled and dialyzed against 1 liter of Buffer B before being loaded onto a Mono Q HR 5/10 FPLC column (Amersham Biosciences) equilibrated with Buffer B. Protein was eluted from the Mono Q column with a linear 0 -0.5 M NaCl gradient (12.5 mM/ml) at a flow rate of 1 ml/min. Mono Q fractions with HsUbc2b activity were pooled and concentrated to less than 5 ml on an Amicon cell fitted with a PM10 membrane and then resolved on a 10 ϫ 30-cm Superose-12 FPLC column (Amersham Biosciences) column equilibrated with 50 mM Tris-HCl (pH 7.5) containing 1 mM DTT and 50 mM NaCl at a flow rate of 1 ml/min.
The isolated HsUbc2b proteins were greater than 99% pure as assessed by SDS-PAGE followed by Coomassie Blue R-250 staining. Absolute protein concentrations were calculated spectrophotometrically using an empirical ⑀ 280 of 1.71 (mg/ml) Ϫ1 for native HsUbc2b, determined by measuring the absorbance at 280 nm of a solution of apparently homogeneous recombinant HsUbc2b for which the absolute protein concentration had been determined by amino acid analysis. Typical final yields ranged from 1 to 1.5 mg of HsUbc2b protein/liter of culture and were usually greater than 90% active by comparing the stoichiometric formation of 125 I-ubiquitin thiolester to the absolute protein concentration. Isolated proteins were stored at Ϫ80°C for several months without loss of activity.
Direct Kinetic Assay of HsUbc2b-125 I-Ubiquitin Thiolester Formation-Initial rates of 125 I-ubiquitin thiolester formation on HsUbc2b were measured in 175-l reactions containing 50 mM Tris-HCl (pH 7.5), 2 mM ATP, 10 mM MgCl 2 , 10 mM creatine phosphate, 1 mM DTT, 10 IU/ml creatine phosphokinase, 5 M 125 I-ubiquitin (about 3,000 cpm/ pmol), 0.7 nM rabbit reticulocyte E1, and the indicated concentrations of HsUbc2b. Aliquots of 25 l were removed at 30-s intervals over the first 1.5 min and quenched with an equal volume of SDS sample buffer. Thiolesters were then resolved from free 125 I-ubiquitin by 12% (w/v) SDS-PAGE under nonreducing conditions at 4°C (29). Bands containing E2 thiolesters, visualized by autoradiography of the dried gels, were excised and quantitated by ␥ counting.
Coupled Kinetic Assay of E1-HsUbc2b Interactions-Binding interactions between the E1 ternary complex and HsUbc2b were quantitated kinetically as the initial net forward rate of E1-catalyzed ubiquitin transthiolation to HsUbc2b by directly coupling this step to E3␣-catalyzed conjugation of human ␣-lactalbumin under E1-limiting conditions. Incubations of 25 l contained 50 mM Tris-HCl (pH 7.5), 2 mM ATP, 10 mM MgCl 2 , 10 mM creatine phosphate, 1 mM DTT, 10 IU/ml creatine phosphokinase, 5 M 125 I-ubiquitin, 0.34 nM rabbit reticulocyte E1, ϳ1.5 g affinity-purified rabbit liver E3␣, and the indicated concentrations of human wild type recombinant HsUbc2b. Samples were incubated for 25 min at 37°C, quenched with an equal volume of SDS sample buffer containing 4% (v/v) ␤-mercaptoethanol, and boiled for 5 min. Radiolabeled conjugates were then resolved from free 125 I-ubiquitin by 12% (w/v) SDS-PAGE and visualized by autoradiography of the dried gels. Radioactive conjugates from each sample were excised from the gel and quantitated by ␥ counting (28).
The interaction of HsUbc2bC88A or HsUbc2bC88S-ubiquitin oxyester with the E1 ternary complex was measured by the concentrationdependent inhibition of wild type HsUbc2b/E3␣-catalyzed 125 I-ubiquitin conjugation of human ␣-lactalbumin under conditions identical to those for the HsUbc2b/E1 assays except that [HsUbc2b] o was held constant at 180 nM. Maximum velocity in the absence of inhibitor was determined in a parallel reaction containing saturating wild type HsUbc2b (1.0 M). The HsUbc2bC88S-ubiquitin oxyester adduct used in the inhibitor studies was stoichiometrically formed during a 30-min incubation at 37°C in a 100-l final volume containing 50 mM Tris-HCl (pH 7.5), 2 mM ATP, 10 mM MgCl 2 , 1 mM DTT, 10 IU/ml PP i , 20 nM rabbit liver E1, 40 M 125 I-ubiquitin, and 34 M HsUbc2bC88S. The resulting HsUbc2bC88S-125 I-ubiquitin oxyester adduct was resolved from other reaction components by gel exclusion chromatography using a 1 ϫ 30-cm analytical Superdex 75 FPLC column (Amersham Biosciences) equilibrated in 50 mM Tris-HCl (pH 7.5) and 50 mM NaCl. The concentration of HsUbc2bC88S-125 I-ubiquitin oxyester was determined from the associated radioactivity of peak fractions quantitated by ␥ counting and confirmed by direct determination of oxyester following nonreducing SDS-PAGE.
Kinetic Assay of HsUbc2b-E3␣ Interactions-Binding interactions between the 125 I-ubiquitin thiolesters of wild type recombinant HsUbc2b and affinity-purified E3␣ were quantitated under E3␣-limiting conditions identical to those in the preceding section with the exception that incubations contained 65 nM rabbit liver E1 and the indicated concentrations of HsUbc2b and E3␣. Inhibition studies were conducted under identical conditions with the exception that wild type [HsUbc2b] o was held constant at 25 nM (HsUbc2bC88A study) or 16 nM (HsUbc2bC88S study), and the concentration of either HsUbc2bC88A or HsUbc2bC88S-125 I-ubiquitin oxyester was varied. Maximum velocity in the absence of inhibitor was determined in a parallel incubation containing 1.0 M wild type HsUbc2b.
Cell Culture-Confluent cultures of IMR90, A549, and Caco-2 cells were maintained at 37°C in a constant atmosphere of 5% CO 2 in Eagle's minimum essential medium supplemented with 5 mM glutamine and 10% fetal calf serum. Two days after the addition of fresh supplemented medium, triplicate 100-mm plates were washed three times with 10 ml of 25 mM phosphate-buffered saline (pH 7.4) and then harvested by scraping into 0.5 ml of ice-cold homogenizing buffer containing 50 mM Tris-HCl (pH 7.5), 0.25 M sucrose, and 1 mM DTT. Cells were then lysed by brief sonication and clarified by centrifugation for 10 min at 16,000 ϫ g. Thiolester assays of a 50-l final volume containing 50 mM Tris-HCl (pH 7.5), 2 mM ATP, 10 mM MgCl 2 , 10 mM creatine phosphate, 1 mM DTT, 10 IU/ml creatine phosphokinase, 5 M 125 Iubiquitin, 68 nM rabbit liver E1, and varying amounts of cell extract were incubated for 5 min at 37°C before quenching with an equal volume of SDS sample buffer from which ␤-mercaptoethanol had been omitted. An aliquot of each sample was adjusted to 2% (v/v) ␤-mercaptoethanol then boiled for 5 min to correct for limited conjugate formation during the incubation. All samples were then resolved from free 125 I-ubiquitin by 12% SDS-PAGE. Bands corresponding to E1-and Ubc2-125 I-ubiquitin thiolester were excised from dried gels and the associated radioactivity quantitated by ␥ counting (29).

E1-catalyzed Formation of HsUbc2b-Ubiquitin Thiolester-
Incubation of the E1 ternary complex with an equimolar concentration of E2 results in rapid stoichiometric transthiolation to form the corresponding E2-ubiquitin thiolester (29,34). In principle, the dependence of initial rate for E2-ubiquitin thiolester formation versus HsUbc2b concentration can be used to examine the kinetics of ubiquitin transfer in the key step linking the half-reactions of ubiquitin activation and ligation under conditions for which [ ratios, the kinetics of transthiolation can be quantitated for the isolated activation half-reaction by measuring the initial rate for approach to equilibrium in the absence of ligase (36). At 0.7 nM E1, initial rates for HsUbc2b-125 I-ubiquitin thiolester formation determined during the first 1.5 min followed hyperbolic kinetics below 0.5 M HsUbc2b, as shown by the linearity of the corresponding reciprocal plot (Fig. 2). Hyperbolic kinetics requires the initial formation of a Michaelis complex between uncharged HsUbc2b and the E1 ternary complex prior to intramolecular transthiolation. Nonlinear least squares fitting of the data in Fig. 2 within the linear region of the reciprocal plot yielded a K m of 102 Ϯ 13 nM for HsUbc2b binding and a k cat of 4.8 Ϯ 0.2 s Ϫ1 , the latter value being calculated as At the highest concentrations of HsUbc2b tested, the initial rates for thiolester formation consistently exhibited substrate inhibition (Fig. 2). This was not a consequence of underestimating v o because rates were linear over at least three time points encompassing the first 1.5 min of the reactions. Linearity during the measurement interval for the kinetic studies also ruled out equilibration of the E1 ternary complex with HsUbc2b-125 I-ubiquitin thiolester observed in Fig. 1, although an equilibrium end point was reached at longer incubation times (not shown). We have subsequently observed analogous substrate inhibition with members of other E2 families (not shown), including the rabbit Ubc8 ortholog E2 20K (37) and the human Ubc4/5 orthologs HsUbc5A/B/C (38). Substrate inhibition of E1-catalyzed transthiolation is consistent with either of two models. In the most plausible model, uncharged E2 can bind two distinct conformations of the E1 ternary complex for which one bound conformer (presumably of lower affinity than the productive binding complex) represents a dead end species. An alternative model in which the E2 carrier protein forms a nonproductive alternate conformation having a low affinity for the E1 ternary complex is much less likely because NMR studies find that the E2 core domain fold is relatively static (39).
Kinetics of E1-HsUbc2b Interactions during E3␣-dependent Conjugation-The previous transthiolation assay monitored rates of HsUbc2b-125 I-ubiquitin thiolester directly. However, the kinetics can also be measured by coupling E2-ubiquitin thiolester formation to the subsequent ligase-catalyzed conjugation step to examine potential effects of ligase on the activation half-reaction. To extend the kinetic studies of E1-HsUbc2b transthiolation, the initial net forward rate of E1-catalyzed HsUbc2b-125 I-ubiquitin thiolester formation was coupled to conjugation of the model N-end rule substrate human ␣-lactalbumin in an E3␣-dependent in vitro assay (23). The step of E1-catalyzed transthiolation is kinetically isolated under conditions for which E1 is rate-limiting, defined by the linearity of v o versus [E1] o and the independence of v o on [E3␣] o . The coupled assay provided the additional advantage of increased sensitivity compared with that of HsUbc2b thiolester formation in Fig. 2. Under E1-limiting conditions, the initial rates of E3␣-catalyzed 125 I-ubiquitin ligation followed normal hyperbolic kinetics with respect to HsUbc2b concentration (Fig. 3). Computer fitting of the data from Fig. 3 yielded a K m of 72 Ϯ 13 nM for uncharged HsUbc2b binding to the E1 ternary complex, in reasonably good agreement with the value of 102 Ϯ 13 nM determined directly in Fig. 2, and an apparent k cat of 4.4 Ϯ 0.2 s Ϫ1 calculated as V max /[E1] o . Correspondence between the kinetic constants determined directly by transthiolation (Fig. 2) and indirectly by the E3␣-coupled reaction (Fig. 3) indicates that the presence of the ligase does not significantly alter either the affinity or catalytic competence of the E1 ternary complex-HsUbc2b interaction. In addition, substrate inhibition by HsUbc2b noted in direct transthiolation assays (Fig. 2) was not observed when this step was coupled to E3␣-dependent ␣-lactalbumin conjugation (not shown), indicating that the former effect was an artifact of the isolated E1-catalyzed half-reaction.
Kinetics of HsUbc2b Transthiolation Catalyzed by Human E1-Recently, Wee et al. (40) reported similar transthiolation kinetics for recombinant human E1 using ubiquitin modified at lysine 6 with the fluorescent label Oregon green. Wee and co-workers report substrate inhibition by HsUbc4 at micromolar concentrations similar to that found for HsUbc2b in the present work (Fig. 2). However, their results differ in several key aspects from those reported here for the activation of ubiquitin by rabbit reticulocyte E1. Because rabbit E1 is 96.5% identical to its human ortholog (41), we extended our transthiolation studies to E1 isolated from human erythrocytes to resolve whether these apparent discrepancies reflected significant differences in catalytic properties of rabbit versus human E1 orthologs, differences in the experimental approaches, or both.
In our hands, human erythrocyte-activating enzyme exhibits the same equilibration between the E1 ternary complex and HsUbc2b-ubiquitin thiolester as found for the rabbit ortholog in Fig. 1 (not shown). As with the rabbit ortholog, we were able to examine the kinetics for transthiolation under initial velocity conditions for approach to equilibrium (36), the results of which are summarized in Table I (Table I) which are statistically indistinguishable from the corresponding values of 102 Ϯ 13 nM and 4.8 Ϯ 0.2 s Ϫ1 found for rabbit E1 (Fig. 2). In contrast, Wee and coworkers report a K m of 1.9 Ϯ 0.5 M and a k cat of 0.07 Ϯ 0.02 s Ϫ1 for HsUbc4 (40).
The nearly 100-fold difference in k cat values for E2 transthiolation suggests that alkylation of ubiquitin with Oregon green at lysine 6 (40) significantly alters the kinetics of transthiolation. This is consistent with the marked difference between wild type ubiquitin and Oregon green-modified polypeptide in their binding to E1 (40) as well as our observation that the values of K m and k cat for HsUbc5A, a member of the same E2 family as HsUbc4 (9), are nearly identical to those of HsUbc2b (not shown). Finally, comprehensive point mutagenesis of ubiquitin has shown that lysine 6 is not required for the initial binding of the polypeptide to human E1 but is essential for subsequent binding of the transition state during formation of the enzyme-bound ubiquitin adenylate. 4 Therefore, derivatization of lysine 6 likely effects a change in rate-limiting step from 4 E. Laleli-Sahin and A. L. Haas, manuscript in preparation. E2 transthiolation to ubiquitin adenylate formation.
Kinetics of E3␣-HsUbc2b-Ubiquitin Thiolester Interactions during E3␣-dependent Conjugation-The rate study of Fig. 3 could additionally be used to probe the step of HsUbc2b-ubiquitin thiolester binding to human E3␣ by kinetically isolating the step of ubiquitin conjugation. This was achieved by increasing the concentration of E1 to 64 nM to make the assay ratelimiting with respect to [E3␣] o , as evidenced by the linear dependence of initial velocity for 125 I-ubiquitin conjugation on [E3␣] o but independence of the observed initial rate on the concentration of E1 (not shown). Under these conditions the concentration dependence of [HsUbc2b] o on the initial rate of human ␣-lactalbumin conjugation is hyperbolic, as demonstrated by the linearity of the corresponding Lineweaver-Burk plot (Fig. 4). Hyperbolic kinetics requires that E3␣ bind the HsUbc2b-125 I-ubiquitin thiolester cosubstrate prior to the catalytic step of target protein-ubiquitin isopeptide bond formation. This conclusion is consistent with earlier qualitative evidence from Reiss et al. (42) for binding of rabbit reticulocyte OcUbc2 to E3␣. Nonlinear least squares fitting of the data from Fig. 4 yielded a value for K m of 54 Ϯ 18 nM, representing the binding of HsUbc2b-125 I-ubiquitin thiolester to E3␣, and a V max of 0.30 Ϯ 0.03 pmol/min. Unambiguous interpretation of the kinetic data in Fig. 4 assumes that the HsUbc2b-125 Iubiquitin cosubstrate for E3␣ be formed stoichiometrically in this E1-coupled reaction. We confirmed the validity of this assumption in parallel incubations resolved by SDS-PAGE under nonreducing conditions for which the steady-state amount of HsUbc2b-125 I-ubiquitin thiolester was stoichiometric with the theoretical amount of carrier protein present in the reaction (not shown). Finally, because there is no analogous stoichiometric assay for functional E3␣, we have no means of reliably estimating k cat from the maximum velocity.
Generation of Nonfunctional HsUbc2b Analogs-Modification of the active site cysteine 88 renders the point mutant HsUbc2bC88A incapable of accepting the ubiquitin thiolester from the E1 ternary complex; therefore, HsUbc2bC88A represents a nonfunctional analog of the uncharged carrier protein (for review, see Ref. 9). In contrast, previous studies show that mutation of the conserved active site cysteine to serine in Arabidopsis thaliana Ubc1 (43) and S. cerevisiae Rad6/ScUbc2 (44) results in E2 polypeptides capable of accepting activated ubiquitin from the E1 ternary complex to form the corresponding E2-ubiquitin oxyester but incapable of supporting subsequent ligase-catalyzed isopeptide bond formation, presumably because of the relative stability of the corresponding ubiquitin oxyester. The analogous ubiquitin oxyester adduct of HsUbc2bC88S is a potential nonfunctional analog of the wild type E2-ubiquitin thiolester. Therefore, these HsUbc2 active site point mutants represent important potential reagents for examining the affinity of the E1 ternary complex for uncharged and charged forms of the ubiquitin carrier protein in carefully designed competitive inhibition assays.
Because HsUbc2bC88A cannot be assayed by functional 125 Iubiquitin thiolester or oxyester assays, the absolute concentration of this point mutant could only be accurately estimated spectrophotometrically in homogeneous preparations using the empirical 280 nm extinction coefficient for wild type HsUbc2b (see "Materials and Methods"). Such quantitation reasonably assumes that the C88A point mutation has no effect on the folding of the polypeptide. To confirm the latter assumption, spectral fitting of circular dichroism data for wild type HsUbc2b and HsUbc2bC88A was employed using the self-consistent method of Sreerama and Woody (45). The percentages of ␣-helix, ␤-sheet, and hydrogen-bonded turn (36,16, and 22%, respectively) were identical between wild type HsUbc2b and HsUbc2bC88A, within the experimental error of the method (45), and comparable with that predicted from the crystal structure of the S. cerevisiae Rad/ScUbc2 ortholog (37, 18, and 21%, respectively) based on Kabasch-Sander secondary structure analysis from the crystal structure of Rad/ScUbc2 (46,47). The excellent agreement between the calculated and predicted secondary structures for HsUbc2bC88A indicates that the mutant exists predominantly in the native form. In addition, convergence of the spectral fit precludes significant contributions from non-native conformers, as discussed previously (45).
The autoradiogram of Fig. 5 (0 min lanes) demonstrates that HsUbc2bC88S is also capable of accepting activated 125 I-ubiquitin from the E1 ternary complex to form the corresponding HsUbc2bC88S-125 I-ubiquitin oxyester adduct. In a parallel rate study, oxyester formation followed first order kinetics with a k o of 1.7 ϫ 10 Ϫ3 s Ϫ1 at 172 nM HsUbc2bC88S and the 20 nM E1 ternary complex, considerably slower than the lower limit of 5 s Ϫ1 estimated for wild type HsUbc2b transthiolation under identical conditions (not shown). Transthiolation is readily reversible for several isozymes, including HsUbc2b, when ATP is depleted, and the E1 ternary complex is quantitatively reversed by addition of excess AMP and PP i (29). In contrast, the HsUbc2bC88S-125 I-ubiquitin oxyester is stable under conditions for which the E1 ternary complex and wild type HsUbc2b thiolesters are readily reversible (Fig. 5, 10 min lanes). The inability to reverse the oxyester indicates that formation of the adduct is irreversible. Other studies confirmed that the HsUbc2b oxyester is incapable of supporting the E3␣-catalyzed ligation of 125 I-ubiquitin to ␣-lactalbumin compared with wild type thiolester (not shown), consistent with earlier studies of 125 I-ubiquitin oxyester adducts of E2 isozymes (9). Therefore, HsUbc2bC88S-125 I-ubiquitin oxyester is catalytically inert once formed and thus meets the criteria for a potential analog of the corresponding thiolester intermediate. The relative stability of the HsUbc2bC88S-125 I-ubiquitin oxyester permitted us to generate reagent quantities of the adduct (see "Materials and Methods") for subsequent use in inhibition studies (below).
Under the conditions of Fig. 5, HsUbc2bC88S formed 72% of  4. Dependence of E3␣-catalyzed conjugation on HsUbc2b concentration. Initial rates for conjugation of human ␣-lactalbumin were determined under E3␣-limiting conditions as described under "Materials and Methods." Incubations were identical to those of Fig. 3 except that the rabbit erythrocyte E1 concentration was increased to 64 nM for each incubation to yield E3␣-dependent initial velocities. the oxyester predicted from the total mutant concentration calculated spectrophotometrically. This value was confirmed by quantitation of the shift in mobility of the oxyester in a parallel gel stained with Coomassie Blue (not shown). The difference between predicted and observed oxyester formation likely reflects the presence of denatured mutant in the preparation; therefore, the active mutant concentration was determined by stoichiometric 125 I-ubiquitin oxyester formation, analogous to the method for quantitating 125 I-ubiquitin thiolester (29).
HsUbc2bC88A and HsUbc2bC88S-Ubiquitin Oxyester Are Competitive Inhibitors of E1-catalyzed Transthiolation-The kinetic study of Fig. 3 (Fig. 6A, closed circles with solid line) from which K i for mutant binding to the E1 ternary complex could be calculated as 104 Ϯ 15 nM. The intrinsic K m determined from the y intercept (85 Ϯ 9 nM) was statistically indistinguishable from that determined in the absence of inhibitor. Good agreement between the value of K i for HsUbc2bC88A binding to the E1 ternary complex and the K m of 102 Ϯ 13 nM for wild type HsUbc2b obtained by direct transthiolation kinetics (Fig. 2) suggests that the structure of the C88A mutant reasonably approximates that of wild type carrier protein. This conclusion is consistent with the comparable CD spectra and calculated secondary structure composition for wild type versus mutant HsUbc2b (previous section). Therefore, the corresponding kinetically determined K m values approximate the equilibrium dissociation constant (K d ) for binding of uncharged HsUbc2b to the E1 ternary complex.
Similar studies using HsUbc2bC88S-ubiquitin oxyester also exhibited competitive inhibition (not shown). Plots of appropriate data according to the modified Michaelis-Menten equation (Equation 1) conformed to the expected linear relationship (Fig.  6A, open circles and dashed line) from which a value for K i ϭ 169 Ϯ 17 nM was calculated. As before, the intrinsic K m determined from the y intercept (119 Ϯ 9 nM) was statistically indistinguishable from that determined in the absence of inhibitor. The data suggest that the E1 ternary complex displays a small but statistically significant affinity preference for binding uncharged HsUbc2b. The difference in binding affinity between HsUbc2b and HsUbc2b-ubiquitin thiolester is much less than the orders of magnitude differences in affinities typical of most substrates versus products. Under some experimental conditions the E2 mutants can potentially serve as competitive inhibitors of the E3␣-catalyzed step used for coupling HsUbc2b-ubiquitin thiolester formation (confirmed below); however, the linearity of the plots in Fig. 6A argue against this possibility under the conditions of the experiment. In addition, separate control studies (not shown) demonstrated that the rate assays remained E1-limiting even at the highest inhibitor concentrations tested in Fig. 6A Fig. 3 but containing 20 nM E1 and 1 M HsUbc2b (wt) or HsUbc2bC88S (C88S) were used to form the corresponding 125 I-ubiquitin thiolester or oxyester, respectively (0 min lanes). Aliquots of these incubations were adjusted to 12 mM 2-deoxyglucose, 10 IU/ml yeast hexokinase, 1.5 mM AMP, and 0.15 mM PP i then incubated an additional 10 min at 37°C to reverse the E1-catalyzed reaction (10 min lanes). Incubations were then analyzed by nonreducing SDS-PAGE and autoradiography (19,29). Migration of the E1 thiolester (E1 SϳUb ), the HsUbc2b-125 I-ubiquitin thiolester/HsUbc2bC88S-125 I-ubiquitin oxyester (HsUbc2b XϳUb ), and free 125 I-ubiquitin are shown on the right. The HsUbc2b thiolester/oxyester migrates as two bands on nonreducing gels because of partial unfolding, as described previously (29). inhibitors had no effect on the steady-state formation of the E1 ternary complex, measured by 125 I-ubiquitin thiolester formation to the activating enzyme (not shown).
HsUbc2bC88A and HsUbc2bC88S-Ubiquitin Oxyester Are Competitive Inhibitors of E3␣-catalyzed Conjugation-When kinetic studies similar to those of Fig. 4 were repeated in the presence of either Ubc2bC88A or HsUbc2bC88S-ubiquitin oxyester, competitive inhibition of the E3␣-limiting step was observed (not shown). Analysis of appropriate data by Equation 1 yielded the predicted linear relationships for the concentration dependence of HsUbc2bC88A (Fig. 6B, closed circles with solid line) and HsUbc2bC88S-ubiquitin oxyester (Fig. 6B, open circles with dashed line) on the initial rates for conjugation. These plots yielded values for K i of 440 Ϯ 55 nM for HsUbc2bC88A and 66 Ϯ 29 nM for HsUbc2bC88S-ubiquitin oxyester. As in Fig. 6A, the K m values calculated from the y intercepts of Fig. 6B were statistically identical to those determined from Fig. 4 in the absence of inhibitor and corresponded to values of 84 Ϯ 18 nM (HsUbc2bC88A study) and 78 Ϯ 14 nM (HsUbc2bC88S study). Separate control experiments confirmed that assays remained E3␣-limiting, as demonstrated by the independence of the initial rates on [E1] o at the highest inhibitor concentrations tested (not shown).
In these experiments, HsUbc2bC88S-ubiquitin oxyester serves as a nonfunctional analog of wild type HsUbc2b-ubiquitin thiolester, the actual cosubstrate for E3␣; in contrast, HsUbc2bC88A mimics binding of the wild type uncharged HsUbc2b product to the ligase. Good agreement between the K i for HsUbc2bC88S-ubiquitin oxyester binding to the ligase (66 Ϯ 29 nM) and the K m for wild type HsUbc2b-ubiquitin thiolester (54 Ϯ 18 nM) indicates that the nonfunctional oxyester serves as a reasonable structural analog of the wild type intermediate. Comparison of these values with the K i of 439 Ϯ 55 nM for binding of the HsUbc2bC88A product analog indicates that E3␣ has nearly a 10-fold discrimination in favor of the thiolester substrate. This requires E3␣ to interact with both the ubiquitin and E2 moieties of the thiolester, although the majority of the binding energy must reside in the carrier protein.
Determination of Cellular E1 and HsUbc2b Concentrations-The previous kinetic results define affinities for protein-protein interactions among the three components of the N-end rule-dependent targeting pathway. To interpret these affinities within a cellular context, the content of endogenous Ubc2 within selected cell types was quantitated in fresh tissue culture cell extracts by its stoichiometric formation of 125 I-ubiquitin thiolester (29), a method used previously to monitor the coordinated induction of E1 and E2 isoforms that accompany the programmed cell death of Manduca sexta intersegmental muscles (48). Extracts were prepared from confluent monolayer cultures and assayed as described under "Materials and Methods." The autoradiogram of Fig. 7 shows typical data from one such thiolester assay in which exogenous rabbit liver E1 (odd numbered lanes) was added to ensure quantitative loading of Ubc2 present in the cell extracts. Separate incubations from which exogenous E1 was omitted were used to quantitate endogenous activating enzyme (not shown). In Fig. 7, thiolesters resolved under nonreducing conditions (odd numbered lanes) are distinguished from the small amounts of conjugates revealed in parallel reducing gels (even numbered lanes) which are formed during the brief incubation (29). Fig. 7 demonstrates that a 125 I-ubiquitin thiolester species comigrating with authentic in vitro formed HsUbc2b-125 I-ubiquitin is the major E2 isoform present in these cells under normal culture conditions, although several other lower abundance E2 thiolester bands are also observed which correspond to E2 EPF (49,50) and HsUbc5/UbcH5 isoforms (38,51), the human orthologs of S. cerevisiae Ubc4/5 isozymes (51) and A. thaliana Ubc8 (52).
Radioactivity associated with thiolester bands for endogenous Ubc2 were quantitated by ␥ counting to calculate the absolute content of each intermediate from the specific radioactivity of 125 I-ubiquitin (29). Values for Ubc2 present in the various cell lines tested, as well as published values from other cell types, are summarized in Table II. The cellular content of Ubc2 varies widely among the cell lines when expressed per 10 6 cells; however, when normalized to cell volume, the resulting concentrations were all within the micromolar range. The content of Ubc2 was consistently lower in erythroid cells (reticulocytes and mouse erythroleukemia cells) than in the other epithelioid and fibroblast-like cell lines (Table II). This observation is consistent with an expanded dependence on alternate targeting pathways in erythroid cells which is inferred from the greater variety and content of other E2 family isozymes in reticulocytes (29,34). The Ubc2 activity reported here for rabbit reticulocytes agrees with earlier estimates by immunological methods (53,54), confirming the validity of Ubc2 quantitation by direct thiolester formation. DISCUSSION In the present studies we have exploited rate measurements as reporter functions for probing protein-protein interactions within a biochemically defined reconstituted N-end rule ligation pathway. Despite the complexity in the overall mechanism for ubiquitin conjugation, such studies demonstrate that the approach provides a facile means of examining this pathway in detail, particularly when combined with genetic manipulation of the interacting partners. In addition, the precision with which the concentrations of E1 and E2 components can be determined by the stoichiometric formation of their respective 125 I-ubiquitin thiolesters provides a means of accurately determining kinetic constants without the need for assumptions regarding the relative content of active protein (19,20). These technical advantages have allowed us, for the first time, to address details of a ubiquitin ligation pathway not otherwise accessible to other techniques.
The rabbit reticulocyte E1 exhibits a relatively high affinity for binding uncharged HsUbc2b and yields a K m of 102 Ϯ 13 nM when measured directly by monitoring the kinetics of transthiolation (Fig. 2). Excellent agreement between the value for K m determined directly and the K i of 104 Ϯ 15 nM determined for the nonfunctional competitive inhibitor HsUbc2bC88A (Fig.  6A) requires that the K m reflect the intrinsic binding affinity (K d ) for the interaction between E1 and the uncharged carrier protein. Side chain interactions between HsUbc2b and its E1 binding site must be largely maintained during the catalytic cycle of transthiolation, with little additional contribution from residues present on the ubiquitin moiety, because the K m for uncharged HsUbc2b approximates the K i of 169 Ϯ 17 nM found for HsUbc2bC88S-ubiquitin oxyester, the isosteric competitive inhibitor of HsUbc2b-ubiquitin thiolester (Fig. 6A). Therefore, there is relatively little difference in affinity (⌬⌬G o ϭ 0.3 kcal/ mol) between the uncharged E2 substrate and ubiquitincharged E2 product of this half-reaction. The nearly identical affinities for substrate HsUbc2b versus product HsUbc2b-ubiquitin thiolester contrasts with most enzymes that exhibit significantly attenuated binding of product as a means of promoting the net forward reaction by favoring dissociation of product. Subsequent product utilization by the E3 ligases appears to drive the E1-catalyzed charging of E2 with ubiquitin thiolester.
In contrast, E3␣ exhibits a marked discrimination in binding affinity between the HsUbc2b-ubiquitin thiolester cosubstrate and the uncharged HsUbc2b product. The K m for HsUbc2b under E3␣ limiting conditions for human ␣-lactalbumin conjugation is 54 Ϯ 18 nM (Fig. 4). Because the empirically determined conditions for these rate studies were set to assure that HsUbc2b was present stoichiometrically as the 125 I-ubiquitin thiolester, this K m reflects the affinity for binding of the corresponding thiolester to the ligase. This interpretation is consistent with the K d of 66 Ϯ 29 nM, measured as K i , for the HsUbc2bC88S-ubiquitin oxyester as a competitive inhibitor of E3␣-catalyzed ␣-lactalbumin conjugation (Fig. 6A), indicating that the K m reflects the intrinsic binding affinity of the wild type E2-ubiquitin thiolester. Prior NMR studies reveal negligible differences between the structures for uncharged and thiolester forms of HsUbc2b (39). Therefore, the ubiquitin moiety of the HsUbc2b thiolester must contribute to binding within the E3␣ active site because the uncharged E2 analog HsUbc2bC88A yields a K d , measured as K i , of 440 Ϯ 55 nM (Fig.  6), representing a nearly 10-fold reduction in affinity (⌬⌬G o ϭ 1.2 kcal/mol). Because HsUbc2bC88A exhibits a K d for binding to E1 which is statistically identical to the K m for wild type HsUbc2b and the secondary structure content of the inactive mutant determined by CD spectroscopy approximates that of the native polypeptide (see "Materials and Methods"), we can preclude significant denaturation of the point mutant as an alternative explanation for the larger K i .
These observations have important implications for correctly interpreting in vivo studies based on overexpression of E2 dominant negative mutants. Frequently, overexpression of ac-tive site C3 A or C3 S mutants of E2/Ubc paralogs has been exploited as a genetic approach for probing the cellular functions of various E2 families (for review, see Ref. 9). Stabilization of short lived proteins following overexpression of specific E2 dominant negative mutants has been interpreted to indicate a role for those E2 isoforms in their ubiquitin-dependent targeting to the proteasome. However, the present studies suggest that the consequences of overexpressing these dominant negative mutants instead potentially reflects indirect inhibition of global E1-catalyzed charging of all E2 isoforms rather than specific E3-dependent effects. Therefore, studies predicated on overexpression of E2 dominant negative mutants should be interpreted cautiously and be accompanied by additional control experiments precluding more general effects of the mutants on global E2 thiolester formation.
In the present studies we have also examined the net forward kinetics of HsUbc2b transthiolation catalyzed by the human ortholog of the ubiquitin activating enzyme (Table I). The K m for ATP and ubiquitin of 7.0 Ϯ 1.1 M and 0.8 Ϯ 0.2 M, respectively, agree favorably with the values of 20 M and 0.9 M, respectively, determined previously in equilibrium studies (55). In addition, the K m values for binding of HsUbc2b by rabbit and human E1 orthologs of 102 Ϯ 13 nM (Fig. 2) and 123 Ϯ 19 nM (Table I), respectively, show remarkable conservation that reflects the high degree of homology between the two enzymes (41). The marked similarity between the rabbit and human E1 orthologs is also reflected in their comparable k cat values for HsUbc2b transthiolation of 4.8 Ϯ 0.2 s Ϫ1 (Fig. 2) and 4.5 Ϯ 0.3 s Ϫ1 (Table I), respectively. The observed k cat for the net forward reaction of rabbit reticulocyte E1-catalyzed HsUbc2b transthiolation is below the empirical lower limit for all internal steps of the E1 catalytic cycle, based on the reported k cat for ATP:AMP exchange of 9 s Ϫ1 (20). This requires that transfer of the ubiquitin thiolester from the E1 ternary complex to HsUbc2b represents the rate-limiting step for HsUbc2b charging rather than an internal step of the E1 mechanism. Therefore, convergence in the k cat values for human versus rabbit E1-catalyzed transthiolation requires a conserved geometry for the transition state for ubiquitin thiolester transfer from the E1 ternary complex to cysteine 88 of HsUbc2b.
Overall, the correspondence between values of K m determined kinetically and K d (determined either directly from earlier equilibrium studies (55) or as K i from inhibitor studies with nonfunctional HsUbc2b mutants) for the three substrates of the E1 reaction indicates that the kinetics of ubiquitin activation approximates a pre-equilibrium rather than the steadystate mechanism proposed earlier based on similar kinetic studies using fluorescently labeled ubiquitin (40). The radioiodinated ubiquitin used in the present studies has been validated previous as functionally equivalent to the wild type polypeptide (19,20); therefore, the apparent discrepancy between the present data and earlier studies of Wee et al. (40) can be best explicated by invoking steric hindrance from the fluorescent label used in the latter work. The marked difference in reported k cat for E2 transthiolation of 4.8 Ϯ 0.2 s Ϫ1 (Fig. 2) versus 0.07 Ϯ 0.02 s Ϫ1 (40) likely reflects a change in the rate-limiting step in the latter study from E1-E2 transthiolation to ubiquitin adenylate formation. A shift to an earlier rate-limiting step because of steric hindrance by the lysine 6-linked fluorescent label would manifest as an overestimation of the K m for binding in subsequent steps. The K m of 1.9 Ϯ 0.5 M for HsUbc4 reported for alkylated ubiquitin (40) versus that of 102 Ϯ 13 nM (Fig. 2) is consistent with the latter prediction, assuming that HsUbc4 binds with similar affinity as HsUbc2b. 5 We have quantitated the content of E1 and Ubc2 within selected cell lines by exploiting the ability of these enzymes to form stoichiometric 125 I-ubiquitin thiolesters (29,48) (Table II). Several general trends emerge from our analysis of E1 and Ubc2 content within the selected cell lines. The Ubc2 isoforms represent the major E2 family present in the four nonerythroid cell lines examined (Fig. 7). Although several of the E2 families have quite similar molecular weights, their corresponding 125 Iubiquitin thiolesters can be resolved by nonreducing SDS-PAGE, allowing their unambiguous quantitation (29,48,56). The intracellular concentration of Ubc2 is consistently within the micromolar range, which is saturating with respect to the K m of 102 Ϯ 13 nM for uncharged HsUbc2b binding to E1 and the K m of 54 Ϯ 18 nM for HsUbc2b-ubiquitin thiolester binding to E3␣. This suggests minimally that Ubc2 exists intracellularly as E1 and E3␣ heterodimers. The data do not address whether stable E1⅐Ubc2⅐E3␣ heterotrimers may also be present; however, we have consistently failed to immunoprecipitate in vitro complexes composed of E1 and E3␣ using an affinitypurified rabbit anti-HsUbc2b polyclonal antibody (not shown). Within IMR90 human lung fibroblasts the concentration of E1 is approximately stoichiometric with Ubc2 and in the range of the concentration of activating enzyme found in rabbit reticulocytes. Near equivalence of E1 and Ubc2 levels in cells appears to hold for other cell lines because we have qualitatively observed similar amounts of their corresponding 125 I-ubiquitin thiolesters in fresh cell extracts (not shown). Rabbit reticulocytes and mouse erythroleukemia cells have atypically low Ubc2 levels (Table II) and differ from the other cell types in possessing significant levels of other E2 families (29,53,57). However, in rabbit reticulocytes the total E2 content is approximately equivalent to that of E1 (29). This suggests a mechanistic requirement for maintaining equivalence between E1 and total E2 levels within cells, presumably to maintain the ubiquitin carrier proteins in their thiolester forms to preclude potential competitive inhibition of the ubiquitin activation and ligation half-reactions observed with HsUbc2bC88A (Fig. 6).
Previous work from our laboratory has quantitated steadystate pools of free and conjugated ubiquitin in various cell lines (for review, see Ref. 58). Comparison of ubiquitin pools with the intracellular content of E1 and Ubc2 suggests that a significant fraction of the "free" ubiquitin is actually present as thiolester intermediates of the ligation machinery. For example, confluent IMR90 fibroblasts contain a total ubiquitin content of 135 pmol/10 6 cells, of which the free ubiquitin represents 54 pmol/ 10 6 cells (58,59). Because IMR90 cells contain 12 and 17.4 pmol/10 6 cells of Ubc2 and E1, respectively (Table II), the actual pool of free ubiquitin is about 7 pmol/10 6 cells, representing an intracellular concentration of about 4 M. The latter value remains saturating with respect to the K m of 0.8 Ϯ 0.2 M for ubiquitin determined with human E1 (Table I). This predicts that significant induction in E2 levels should be accompanied by a coordinated induction of ubiquitin to avoid depletion of free intracellular ubiquitin pools.
The present studies provide the first comprehensive examination of protein-protein interactions within a ubiquitin ligation pathway and the relationship between the resulting binding constants and intracellular concentrations of the targeting components. The results demonstrate that detailed features of the ubiquitin ligation mechanism can be resolved by appropriately designed rate studies.