The Mechanism of Ubiquitin Activating Enzym.e A KINETIC AND EQUILIBRIUM ANALYSIS*

The recently characterized ubiquitin activating en- zyme catalyzes the first step in ubiquitin-protein isopeptide bond formation and as such may be central to a number of regulatory events in addition to its previously reported role in energy-dependent proteolysis. The present work substantiates the following sequence of reactions in ubiquitin activation: 1) initial formation of tightly enzyme-bound ubiquitin adenylate with PPi formation from ATP; 2) conversion of this intermediate to form AMP and a covalent enzyme-ubiquitin thiolester; and 3) activation of a second molecule of ubiquitin to give a ternary complex of one equivalent each of ubiquitin thiolester and tightly bound ubiquitin adenylate per subunit of enzyme. Isotope exchange kinet- ics at equilibrium demonstrate that substrate binding and product release are both strictly ordered, with ATP the leading substrate with respect to ubiquitin and PPi the leading product with respect to AMP. Equilibrium constants relating all resolved steps of the mechanism have been determined by direct measurement of ubiquitin adenylate and ubiquitin thiolester formation. The equilibrium constant for formation of enzyme- bound ubiquitin adenylate and PPi from bound substrates is 0.16. Ubiquitin distribution between the bound adenylate and the enzyme-ubiquitin thiolester plus bound AMP favors the latter by a factor of 2. Reactant dissociation constants were found to be 0.45 CM for ATP, 0.58 PM for ubiquitin, and 3.2 PM for PPi. Free enzyme shows no apparent binding of AMP, but the ubiquitin thiolester form of enzyme binds this mon- onucleotide with a dissociation

(4) with the carboxyl terminal glycine of ubiquitill to form an isopeptide bond ( 7 ) identical with that demonstrated for the histone conjugate A24 (8). In reticulocytes such conjugates are committed to degradation (6) with the subsequent liberation of free ubiquitin (4). In other cells ubiquitin conjugation to nucleosomal core histones 2A (8) and 2B (9) may serve to regulate chromatin structure during transcription (for a recent review see Ref. 10). Thus, ubiquitin ligation represents a novel form of protein modification that may function in multiple regulatory modes in addition to its role in protein breakdown.
The enzyme responsible for the activation of ubiquitin, purified to homogeneity (ll), is a dimer comprised of 1 X lo5 dalton subunits. The enzyme exhibits both ubiquitin-dependent ATP:PPi and ATP:AMP exchange (11,12). In the presence of ATP a covalent enzyme-bound ubiquitin is formed having properties consistent with a thiolester between the carboxyl terminal glycine of the polypeptide and a sulfhydryl residue of the enzyme (7, 12). Conjugate bond formation proceeds from this ubiquitin thiolester, in the absence of ATP, by a subsequent enzyme of the system (13). Therefore, in addition to being the first enzyme of the degradative pathway, the ubiquitin activating enzyme probably represents the only energy-requiring step in proteolysis, although other fates of activated ubiquitin may make the ubiquitin activating enzyme central to a variety of cellular functions. Ubiquitin activating enzyme catalyzes the biphasic ubiquitin-dependent formation of PPI from ATP. This process consists of a rapid pre-steady state release of two equivalents of PP, per subunit of enzyme followed by a significantly slower linear release of PPi, the latter process being proportional to both the concentrations of enzyme and thiols such as dithiothreitol (13). The two equivalents of ATP hydrolyzed during the pre-steady state result in the formation of one equivalent each of covalently enzyme-bound ubiquitin thiolester and a second species of ubiquitin non-covalently bound to enzyme, ubiquitin adenylate (13). Knowing that the carboxyl-terminal glycine of ubiquitin is involved in the thiolester linkage to the enzyme (7), the ubiquitin adenylate is believed to be a mixed acyl phosphate anhydride between the carboxyl terminus of ubiquitin and AMP (13). A minimal mechanism for the formation of a ternary complex composed of the two forms of enzyme-bound ubiquitin has been proposed to consist of three steps (13): E$Y''h e E s I I~ + AMP (2) E s ,~h + ATP + Ub + E?P/x'"'" + PP, The product of step 3 has been isolated and shown to have the expected ratio of 2:1 for enzyme-bound ubiquitin:AMP The Sequence and Distribution of Enzyme Intermediates per subunit (13). The enzyme must possess an exceptional affinity for ubiquitin adenylate since, once formed, this intermediate does not measurably dissociate. The mechanism explains the finding that ATP:PP, exchange occurs in the absence of added AMP, probably entirely by way of step 3, but that ATP:AMP exchange requires PP, since the latter reaction must encompass both steps 1 and 2 (12). The great stability of the enzyme-products complexes makes the ubiquitin activating enzyme inaccessible to a steady state kinetic examination. Isotope exchange at equilibrium can be used to quantitate the flux through the three steps and establish the order of reactant participation. Such studies also allow one to test the principal feature of the minimal model: the existence of two forms of chemically exchangeable enzyme-bound ubiquitin. As will be shown, it is additionally possible to determine the equilibrium constants for the steps of the scheme, when expanded t o include all Michaelis complexes, by measuring the distribution of these two forms of enzyme-bound ubiquitin under a range of equilibrium conditions. Observation of unusually tight binding for AMP ( K dlo-" M ) indicates that this mononucleotide may play an important physiological role in regulating the rate of ubiquitin conjugate formation by determining the distribution of enzyme-bound forms in Step 2.

MATERIALS AND METHODS
Ubiquitin was purified to homogeneity from human erythrocytes by an extension of previous methods (14,15). Concentrations of ubiquitin stock solutions were measured relative to a standard determined by amino acid analysis (5) and calculated based on a molecular weight of 8565. A portion of the purified ubiquitin was labeled with I2'I by the chloramine-T method ( 3 ) . The functional equivalence of native and iodinated ubiquitin for the activating enzyme has been demonstrated previously by the good agreement found in stoichiometry studies using '"I-ubiquitin, [y-"'P]ATP, or [:'H]ATP (13). Carrierfree NaI2'I was obtained from Amersham; [ynZP]ATP, [G-"HIAMP, [2,8-"HIATP, and Nar'"PP, were obtained from New England Nuclear.
Reticulocyte-rich whole blood was obtained by phenylhydrazine induction of adult male rabbits (16) and used to prepare fraction I1 as previously described (1). Ubiquitin activating enzyme was purified to apparent homogeneity from fraction I1 by covalent affinity chromatography, adjusted to 0.1 mM dithiothreitol and 1 mg/ml of bovine serum albumin to enhance stability, then extensively dialyzed as described (13). The enzyme was divided into small aliquots and stored at -80 "C. Although the enzyme retains full activity for at least 8 months when stored at -80 "C in the presence of dithiothreitol and bovine serum albumin, a slight loss of activity is observed with repeated freezing and thawing. Therefore, all experiments were performed with aliquots thawed only once. Activating enzyme was quantitated by the extent of ubiquitin [,'H]adenylate formation in the presence of [:'H]ATP and pyrophosphatase (13). formed a t 37 "C in a final volume of 50 pl containing 50 mM Tris-C1, Isotope Exchange Assays-All isotope exchange assays were per-pH 7.6, 10 mM MgCI2, 0.1 mM dithiothreitol, and 0.27 pmol of activating enzyme. Concentrations of ATP, AMP, PP,, and ubiquitin were as described in the accompanying figure legends. Incubation times were chosen to measure only initial exchange velocities within the linear portion of the progress curve and, therefore, required no correction for approach to equilibrium.
For ATP:PP, exchange, the incorporation of "2PP, into ATP was determined by adsorbing the latter onto charcoal. Reactions were quenched by addition of 0.5 ml5% (w/v) trichloroacetic acid containing 4 mM carrier PP, followed by 0.3 ml of a lo%> (w/v) slurry of charcoal in 2% trichloroacetic acid. After centrifugaticn the charcoal pellet was washed with 3 X 1 ml 2% trichloroacetic acid and bound radioactivity determined directly by Cerenkov radiation (13). Data were corrected for a control assay from which enzyme had been omitted and that generally represented <0.2% of added radioactivity (4-6 x 10' total cpm). Assays were performed in duplicate and consistently agreed within 5% of their mean value. No exchange was observed in the absence of ubiquitin (not shown).
For ATP:AMP exchange, reactions were quenched with 55 pl of 0.2 M potassium phosphate buffer, pH 2.0, containing 0.3 M MgCI2, and the incorporation of ['HIAMP into ATP was determined by high performance liquid chromatography separation of the nucleotides on a Varian 5000 instrument equipped with an AX-10 anion exchange column (Varian) equilibrated with 0.1 M potassium phosphate buffer, pH 2.0, and 0.15 M MgCL Nucleotide peaks were collected and quantitated by liquid scintillation counting. Data were corrected for a control assay from which enzyme had been omitted and that represented -0.8% of the added radioactivity (-10" total cpm). By this procedure, reproducibility was such that duplicates consistently agreed within 5% of their mean value. No measurable exchange was observed in the absence of either PP, or ubiquitin (not shown).
Measurement of Enzyme-bound Ubiquitin Adenylate a n d Ubiquitin Thiolester-All studies measuring the equilibrium concentrations of ubiquitin adenylate or ubiquitin thiolester were conducted at 37 "C in 50 mM Tris-C1, pH 7.6, containing 0.1 mM dithiothreitol and 10 mM MgC12. Concentrations of ATP, ubiquitin, PP,, and AMP were as described in the accompanying figure or table legends.
Enzyme-bound ubiquitin adenylate was determined by acid-precipitable radioactivity using ["HIATP (13). Reactions of 50 p1 of final volume containing 0.8 mg/ml of bovine serum albumin were initiated with enzyme then incubated 4 min and quenched by addition of 100 pI of ice-cold 16% trichloroacetic acid. After standing on ice for 10 min, protein in the sample was sedimented by centrifugation at 15,000 X g for 10 min. The resulting pellet was rinsed twice with cold 2% trichloroacetic acid then dissolved in 100 p1 of 0.2 M Tris-C1, pH 7.6, and "H radioactivity determined by liquid scintillation. Data were corrected for a control incubation from which enzyme had been omitted and which generally represented -0.2% of total added radioactivity.
Enzyme-bound ubiquitin thiolester was determined using '"I-labeled ubiquitin. Reactions of 20 pl of final volume were initiated with enzyme, then incubated 5 min and quenched by addition of 20 p1 of 4 M urea, 4% NaDodS0,-PAGE sample buffer. Covalently enzymebound '2,51-ubiquitin thiolester was resolved from free '"I-ubiquitin by NaDodS0,-PAGE as described previously (13), except that the gel was dried and autoradiographed without staining. The band corresponding to enzyme was cut from the dried gel and radioactivity quantitated in an Abbott Autologic gamma counter. Data were corrected for a control to which enzyme had been introduced after addition of 4 M urea, 4% NaDodSO, sample buffer and which was consistently less than 0.1% of total added radioactivity. The reproducibility of this method for quantitating ubiquitin thiolester was such that independent replicates consistently agreed within 5% of their mean value.

Equilibrium Isotope Exchange Studies
AMP Dependence of ATP:PPi a n d ATP:AMP Exchange-The ATP:PPi exchange rate at equilibrium exhibits an unusual response to changes in AMP concentration, as shown in Fig. 1. At either low or high concentrations, pyrophosphate exchange is independent of AMP. That no response of ATP:PP, exchange to AMP is observed at these extremes of concentration is consistent with the absence of an AMP requirement for this process (12). However, within a limited range of concentration an effect of AMP on the exchange rate is observed. The shape of the curve within this sensitive region of AMP concentration suggests a titration-like process. The inflection point defining the concentration of mononucleotide required for half maximal change in rate, K I~, is inversely related to the concentration of PP, in the assay. The K I ,~ value for AMP increases from 8 PM at a PP, concentration of 100 PM to a value for K I~ of 20 PM at 16 PM PPi (Fig. 1). This effect of AMP is seen only in the pyrophosphate exchange reaction, since the AMP dependence on ATP:AMP exchange displays normal hyperbolic kinetics as shown by the linearity of a double reciprocal plot for such data in Fig. 2. At 10 pM PPi the maximal velocity of ATP:AMP exchange yields an enzyme turnover rate of 1.3 s" with a K l p for AMP of 100 PM.
The apparent partial inhibition by AMP in Fig. 1 cannot be due to competition by AMP for either ATP or PPi binding since the ATP:PP, exchange rate does not continue to decrease at high concentrations of the mononucleotide. The latter observation also rules out formation of an abortive E e AMP complex as does the linearity of the reciprocal plot in Fig. 2 under similar conditions. However, the AMP dependence of ATP:PPi exchange is consistent with a shift between two parallel ATP:PP, exchange pathways that depends on binding of AMP to the enzyme. Such parallel ATP:PP, exchange pathways likely represent steps 1 and 3 of the proposed mechanism. A shift between these two steps would then depend on the position for the equilibrium represented by step 2 as influenced by AMP.
To test this directly, the extent of enzyme-bound "'I-ubiquitin thiolester formation was determined after resolution from free '"I-ubiquitin by NaDodS04-PAGE (13). When 1.16 pmol of enzyme is incubated in the presence of 1 mM ATP, 10 mM MgC12, 100 PM PP,, and 1.2 pM "'1-ubiquitin, one finds an equivalent amount of labeled ubiquitin covalently bound to enzyme as thiolester a t equilibrium after correction for a control sample (Experiment 1, Table I). When an identical incubation is conducted in the presence of 2 mM AMP, only 0.05 pmol (4.3%) of thiolester is formed. This effect of AMP on the equilibrium concentration of enzyme-bound ubiquitin thiolester is not observed if the nucleotide is added to the incubation after first quenching with urea-NaDodS04, precluding any nonenzymic artifact. Therefore the equilibrium position of step 2 lies far in favor of thiolester formation at the level of AMP generated by the reaction and added nucleotide shifts the exchange process from step 3 to step 1.
That the simultaneous presence of PP, is required for AMP to achieve this effect is shown when a similar experiment is conducted in the presence of inorganic pyrophosphatase to remove PPi generated in the reaction and allow quantitative formation of the ternary complex (13), Experiment 2 of Table I. In the absence of PPi, AMP at 2 mM has no effect on the extent of ubiquitin thiolester formation. This observation, that with ubiquitin adenylate present on the enzyme AMP is unable to dislodge the ubiquitin thiolester, suggests that there is only one site for binding either AMP or ubiquitin adenylate. ATP Dependence ofATP:PP, a n d ATP:AMP Exchange-Shown in Fig. 3 are double reciprocal plots for the ATP dependence on the rates of ATP:PPi and ATP:AMP exchange.
[ATP]", pM" TABLE I Equilibrium formation of enzyme-bound ubiquitin thiolester Incubations were performed as described under "Materials and Methods" in the presence of 1 mM ATP, 1.2 p~ '"I-ubiquitin (1890 cpm/pmol), and 1.16 pmol of activating enzyme. Experiment 1: incubations contained 100 p~ PP, in the absence or presence of 2 mM AMP. As a control, AMP was added to an identical sample after quenching with 4 M urea, 4% NaDodSOI sample buffer as indicated. Experiment 2: incubations were identical to Experiment 1 but with 0.87 pmol of enzyme and 0.15 IU of inorganic pyrophosphatase in place of PP, in the incubation.

Radioactivity
Ubiquitin thiolester The apparent maximum enzyme turnover rates for PP, and AMP exchange are 4.9 and 1.1 s", respectively. The concentration of ATP required to obtain half-maximal velocity for PP, exchange is 36 PM, a value in good agreement with a K J I 2 of 38 PM for AMP exchange. The data in Fig. 3 alone do not reveal the sequence of substrate binding to activating enzyme. However, the data allow one to preclude the existence of any types of abortive E. ATP complexes.
Ubiquitin Dependence of ATP:PP, a n d ATPrAMP Exchange-The dependence of both PP, and AMP exchange on ubiquitin is shown in Fig. 4. For ATP:PPi exchange, the effect of ubiquitin was tested in the absence or presence of 1 mM AMP to shift the observed exchange segment from step 3 to step 1, respectively. Ubiquitin at concentrations above 2 PM (17 pg/ml) causes severe inhibition of pyrophosphate exchange by either step 1 or step 3. Similar inhibition by ubiquitin is observed for ATP:AMP exchange. Within the region of inhibition, increasing ubiquitin concentration causes the initial velocities of both PP, and AMP exchange to ap- proach zero, indicating a n absolutely ordered addition of substrates with ubiquitin binding to enzyme after ATP.
That ubiquitin causes an identical effect on PP, exchange by either step 1 or step 3 is reasonable if formation of ubiquitin adenylate occurs at the same site in both of these reactions. Since ATP:AMP exchange must proceed through step 1, one would expect the ubiquitin dependence of this exchange to mirror that of PP, exchange by the same step. Double reciprocal plots of data from Fig. 4 are linear for ubiquitin concentrations below 1-2 ~L M (not shown) and allow estimates of apparent maximal turnover and KlI2. For ATP:PPi exchange in the absence and presence of 1 mM AMP, the maximal turnover rates are 9.6 and 5.6 s-' and the K1,* values are 1.2 and 0.6 p~, respectively. For ATP:AMP exchange, the values are 1.9 s-l for maximal turnover and 0.7 p~ for the Kl/z of ubiquitin.

PP, Dependence of ATP:PP, and ATP:AMP Exchange-
The concentration dependence of PPi on both ATP:PPi and ATP:AMP exchange was tested to determine the order of product release from the enzyme. The exchange of PPi into ATP via step 3, involving the ubiquitin thiolester form of enzyme, displays a normal hyperbolic dependence on PPi (Fig.  5 ) , and yields a linear double reciprocal plot (not shown). The maximal turnover rate is 6.2 s-' and the KIl2 for PPi exchange is 5 p~ from such a reciprocal plot. In contrast, PPi is found to be a product inhibitor of ATP:AMP exchange above 10 PM (Fig. 5 ) . Below 10 p~ a double reciprocal plot is linear (not shown) allowing estimations for maximal turnover and K1/2 of 1.0 s-l and 0.7 p~, respectively. With increasing concentration of PPi the velocity of AMP exchange tends to zero. That PPi is not similarly a product inhibitor of ATP:PP, exchange when catalyzed by step 1 was shown earlier by the rates for this reaction at 16 and 100 p~ PP, for high concentrations of AMP, Fig. 1. These observations indicate an ordered release of products from the: enzyme in which the PPi formed in step 1 dissociates prior to the release of AMP generated in step 2.

Equilibrium Constant Determinations for Individual Steps
Based on previous studies of stoichiometry (13) and on the equilibrium isotope exchange experiments in the present work, an expanded minimum mechanism including aU Michaelis complexes may be given by Scheme I for the ubiquitin activating enzyme.* Supporting evidence for Scheme I is sum-' Scheme I neglects the required role of Mg'+ since the present studies did not hear directly on this question. However, by analogy to similar mechanisms, it is reasonable to assume the actual substrates involved are ATP. Mg'+ and PPI. Mg'+. marized under "Discussion." As is usually the case in all but the simplest mechanisms, Scheme I is too complex to allow evaluation of individual dissociation or equilibrium constants by kinetic methods alone. However, having identified the principle steps and their sequence, the constants identified in Scheme I could be determined graphically by measurement of ubiquitin adenylate and ubiquitin thiolester present over a range of equilibrium conditions.
Determination of KL1 and KL-Ubiquitin adenylate concentration as measured by acid precipitation of ubiquitin ["HI adenylate ( When the ubiquitin dependence for formation of the adenylate intermediate is determined at fixed concentrations of ATP and the data plotted according to Equation 4, one obtains a set of lines having a constant intercept but slopes inversely related to [ATP] as predicted, Fig. 6. If the slopes of these lines are plotted uersus the reciprocals of their respective ATP concentrations (Fig. 6, inset) K l l may be evaluated as the ratio of the s1ope:intercept of the secondary plot. One obtains a value of 0.45 p~ for K'-l, the dissociation constant for ATP from the thiolester form of enzyme. From the intercept values for the primary and secondary plots of Fig. 6, one may calculate a value of 0.58 p~ for the dissociation constant for ubiquitin, Klp. In preliminary studies it was found that the data remained linear when plotted according to Equation 4 for ubiquitin concentrations up to 12 p~ at the ATP concentrations used in Fig. 6. Thus, the substrate inhibition by ubiquitin observed for isotope exchange in Fig. 4 is not reflected in the equilibrium formation of ubiquitin adenylate. This observation precludes schemes involving random binding of ATP and ubiquitin for which the initial binding of ubiquitin, rather than ATP, results in an abortive complex for isotope exchange.
Determination of K':] and KL-At saturating ubiquitin, Equation 4 reduces to a form that can be rearranged to yield  Fig. 7 , K'j is directly determined at 0.16 while the ratio of s1ope:intercept yields a value of 3.2 IJM for KL,.
Determination of K-s-The stability of enzyme-bound ubiquitin thiolester to isolation by NaDodS04-PAGE, as illustrated in Table I, provided a method for quantitating the concentration of this enzyme form. In Table I   The Sequence and Distribution of Enzyme Intermediates 6 (see "Appendix" for derivation and definitions for / I I and p2) a linear plot is observed as shown in Fig. 8.
The ratio of s1ope:intercept yields a value for ,f32K-6 of 200 p~. As defined in the "Appendix," p 2 contains only known concentrations and experimentally determined equilibrium constants. Substituting these values into the expression for ,&, K-$ can be calculated as 0.027 p~. In contrast, the apparent KIr2 value for AMP with respect to ubiquitin thiolester formation is -10 p~ (Fig. 8, inset). This latter value is in good agreement with the K1p of 8 p~ determined from the shift in ATP:PPi exchange between step 1 and step 3 at the same concentration of PP, (Fig. 1).
Determination of K,-The equilibrium constant for ubiquitin transfer between adenylate and thiolester forms of enzyme, K,, was evaluated in a similar manner by measuring the effect of PPi on the concentration of thiolester at a constant level of AMP. The dependence of thiolester formation on PP, is described by Equation 7 (see "Appendix") from which Ks can be determined from the ordinate intercept." At [PP,] > KL4, Equation 7 predicts the value for the complex term p2 to be independent of pyrophosphate and approach the value given by Fig. 8 determined at 100 p~ PPi. That a plot of appropriate data according to Equation 7 is linear (Fig. 9) rather than concave up demonstrates the validity of the latter prediction. A value of K, equal to 2.0 was calculated from the ordinate intercept of Fig.   9 using a value for ,& K-ti of 200 p~ obtained from Fig. 8.

DISCUSSION
A minimal model for the ubiquitin activating enzyme, given in steps 1-3, was proposed previously on the basis of stoichiometry studies that indicated that the utilization of two equivalents of ATP per subunit of enzyme resulted in the formation of one equivalent each of enzyme-bound ubiquitin adenylate and ubiquitin thiolester (13). The model accounts for the observed properties of two ubiquitin-dependent exchange reactions: an ATP:PP, exchange that occurred in the absence of added AMP and an ATP:AMP exchange that required the presence of PPi (12). This work was initiated to provide additional evidence for the minimal model, establish in detail the sequence of interactions, and determine the thermodynamic and kinetic behavior of the intermediates.
From the kinetic data presented in this work the minimum mechanism for activating enzyme can be expanded to include all Michaelis complexes as shown in Scheme I. The strict order of substrate binding is indicated by the normal hyperbolic kinetics with varying ATP (Fig. 3), but inhibition by ubiquitin of exchanges that involve ATP (Fig. 4). Similarly, the ordered release of PPi and AMP follow from the normal hyperbolic kinetics for varying PPi in ATP:PP, exchange but I Equation 7 could be further simplified by conducting the experiment at [AMP] > ,f?,K-I;. However, this would have introduced error into the ordinate intercept determination because of the resulting increased slope. For this reason, an intermediate concentration of [AMP] equal to 0.5 of the value for BaK 6 was chosen in this experiment. Calculation of K., from an ordinate intercept allowed one to avoid the a priori assumption of equivalence between constants determined for the thiolester form of enzyme and those for enzyme in which this site was unoccupied. PP, inhibition of ATP:AMY exchange (Fig. 5). The linearity of double reciprocal plots for isotope exchange studies in Figs. 2 and 3 provide no evidence for abortive complexes involving either AMP or ATP, respectively. While the substrate inhibition observed for the ubiquitin dependence on isotope exchange in Fig. 4 does not alone rule out formation of abortive enzyme-ubiquitin complexes, this possibility does not seem likely since ubiquitin at up to 12 p~ shows no similar inhibition of equilibrium ubiquitin adenylate formation as noted earlier.
By analogous reasoning the linearity of data for the effect of PP, on the equilibrium concentrations of ubiquitin adenylate (Fig. 7) and ubiquitin thiolester (Fig. 9) preclude formation of any abortive complexes involving PP, that would not have been resolved by the concentration dependence of this reactant on isotope exchange between ATP and AMP (Fig. 5 ) .
Scheme I is further substantiated by the observation that AMP affects the rate of ATP:PP, exchange within a critical concentration range but is without effect above or below this region (Fig. 1). The effect of AMP on the rate of ATP:PP, exchange is best explained by a shift of this exchange process from one occurring on enzyme forms involving E~. w , to forms involving Est$ by perturbation of tne equilibrium position for enzyme-bound ubiquitin thiolester. This interpretation is demonstrated by three observations: 1) the transition between the two limiting AMP-independent rates of ATP:PPi exchange is accompanied by a loss of enzyme-bound ubiquitin thiolester at high nucleotide concentration (Experiment 1, Table I); 2) the effect of AMP concentration on the rate of ATP:PPi exchange and on the extent of ubiquitin thiolester formation show good correspondence as judged by the nearly identical values for Klrz of 8 p~ (Fig. 1) and 10 p~ (Fig. 8, inset), respectively, for the two processes under the same conditions; and 3 ) the concentration of AMP giving half-maximal change in ATP:PPi exchange, and ubiquitin thiolester concentration by inference, is itself inversely related to the concentration of PPi (Fig. 1). The last point is a consequence of the individual sites for ubiquitin thiolester and ubiquitin adenylate shown from stoichiometry studies of the ternary complex, in addition to the extremely tight binding of the adenylate intermediate (13). While PPi would not be expected to participate directly in determining the equilibrium concentration of ubiquitin thiolester, it is required for the effect of AMP on this step to be observed (Experiment 2, Table I) and does itself affect the equilibrium position for ubiquitin adenylate formation (Fig.   7 ) . The influence of PPi on thiolester concentration reflects the mutally exclusive binding of AMP, ubiquitin adenylate, and ATP to a single site on E S -U H .
Thus, one would predict PP, to act in concert with AMP in defining the concentration of total ubiquitin thiolester.
Previously, ubiquitin thiolester was demonstrated as the immediate donor for protein-conjugate formation (13). That the concentration of AMP directly and of PP, indirectly, determine the equilibrium position for formation of this activated ubiquitin intermediate, suggests product inhibition as a possible regulatory mechanism for the rate of conjugate formation and ultimately protein degradation. The previous report by Hershko et al. (17) that AMP was a potent inhibitor of energy-dependent proteolysis in reticulocyte extracts is consistent with this hypothesis. With a dissociation constant of 0.027 p~, the AMP concentration of 10 to 100 p M in cells (18,19) seems sufficient totally to suppress ubiquitin thiolester formation and ultimately energy-dependent proteolysis. Concentrations of AMP much greater than that required for apparent saturation are necessary for inhibition of in vitro proteolysis (17). Assuming the rate-limiting step for this degradative pathway occurs subsequent to ubiquitin activation, the latter observation suggests PPi regulates the activating The Sequence and Distribution of Enzyme Intermediates 10335 enzyme by determining the K I l 2 a t which AMP influences the distribution between forms involving E s . u~ and E S H . Therefore, the low intracellular concentrations of PPi one might anticipate allow the enzyme to function in the presence of an intracellular AMP concentration that would otherwise cause complete inhibition. Since AMP and PPi are products of many key anabolic reactions, product inhibition by these metabolites at the level of ubiquitin activating enzyme may coordinate cell growth with proteolysis as has been noted in the decreased rate of protein degradation during liver regeneration (20,21).
Factors that contribute to the high affinity of the enzyme for AMP may also be responsible in part for the observed tight binding of ubiquitin adenylate to the enzyme (13). There are indications that the enzyme is able to exploit this inordinately tight association of ubiquitin adenylate to normalize the equilibria involved in its formation and utilization. From the standard free energies a t 25 "C, pH 7.0, for hydrolysis of ATP to yield AMP + PPi (-7.7 kcal/mol; Ref. 22) and of acetyl adenylate to give AMP + acetate (-13.3 kcal/mol; Ref.  (Table II), or a relative destabilization of this intermediate by -5.4 kcal/mol. The opposing directions and similar magnitudes of these thermodynamic differences suggest that the enzyme couples the highly favorable binding energy for ubiquitin adenylate to the unfavorable equilibrium constant for the formation of this intermediate. The thermodynamic advantage gained by increasing K's would be achieved at the expense of diminishing the stability of the enzyme-bound ubiquitin thiolester. However, such an affect may itself serve a selective advantage by enabling regulation of the enzyme via product inhibition. One might test whether the binding energy of ubiquitin adenylate is coupled to the normalization of these equilibria since K':3 and K, should be related to the observed dissociation constants for mononucleotides structurally related to AMP. One would predict the value of K's to decrease and K5 to increase with diminishing affinity for a series of such alternate substrates.
Ubiquitin adenylate formation represents but one example of a much broader class of enzymic reactions involving car-boxylate activation via adenylation. The most pertinent analogy in the present context are the aminoacyl-tRNA synthetases which catalyze nucleotidyl transfer between enzymebound ATP + amino acid to form aminoacyl adenylate + PP,.
How closely does the equilibrium constant for aminoacyl adenylate formation correspond to the chemically analogous step for ubiquitin adenylate formation defined by the constant K':$ (or K13)? Such a constant has been measured only twice previously. For Escherichia coli methionyl-tRNA synthetase the equilibrium constant for nucleotidyl transfer in the direction of methionyl adenylate formation has been determined kinetically as 1.5 at 25 "C, pH 7.6 (24), and has recently been confirmed by direct measurement using "'P NMR (25). Kinetic studies of E. coli isoleucyl-tRNA synthetase yield a value of 0.2 for the analogous nucleotidyl transfer a t 25 "C, pH 8.0 (26). The value for K r a of 0.16 shows surprisingly good agreement with these chemically similar but functionally diverse reactions. The correspondence in these values presents the intriguing alternatives of either convergent evolution toward a step of maximum catalytic efficiency or the divergent evolution of the ubiquitin activating enzyme from an ancestral aminoacyl-tRNA synthetase. In either case, it is likely that aminoacyl adenylate binding on the synthetases is also coupled to the stabilization of this intermediate. Experimentally the ubiquitin activating enzyme provides a better system for testing the validity of this hypothesis since the equilibria are more conveniently measured and ubiquitin transfer to form thiolester provides a good thermodynamic model for enzymebound aminoacyl adenylate transfer to tRNA. The method of measuring ubiquitin [3H]adenylate did not lend itself to direct determination of the individual constants comprising the upper reaction path in Scheme I. The high concentration of AMP required to shift the equilibrium position of enzyme-bound intermediates to the upper path would have led to experimental problems associated with isotope exchange and dilution. One might ask if occupancy of the thiolester site with ubiquitin influences the distribution of intermediates at the adenylate site. Three lines of evidence suggest that there is a minimal interaction of this type between the two ubiquitin sites: 1) the limiting rates of ATP:PP, exchange do not differ greatly a t low or high concentrations of AMP (Fig. 1); 2) the Kl/z values of ATP for ATP:PPi and ATP:AMP exchange are comparable (Fig. 3); and 3) the estimated K1/2 values of ubiquitin for ATP:AMP exchange and for ATP:PPi exchange in the absence and presence of 1 mM AMP show good agreement with values of 0.7,l.Z and 0.6 pM, respectively. Therefore, it appears that the equilibrium and dissociation constants summarized in Table I1 approxi-  (Fig. 2) one may determine analogous constants for AMP.
As summarized in Table 11, the bimolecular association constants for ATP and PPi show good agreement. This observation may indicate that a common conformational change limits the rate of ATP and PPi binding. Such a conformation change would be consistent with the strictly ordered mechanism of Scheme I for which no evidence for abortive complexes was observed. The tighter binding of AMP relative to ATP results from a greater associative rate constant for AMP of 7.9 X 10' M-' s-' that approaches the diffusion limit. Although this rate constant for binding of AMP to ES.Ut, is 20-fold greater than that for ATP, the respective dissociation rate constants for the two nucleotides to yield this enzyme form are nearly identical. The comparable values for kP6 and kL1 may reflect a common rate-limiting conformational relaxation of the protein involved in dissociation of either nucleotide to yield E S -U b . Further, this suggests that the difference in nucleotide binding rates is due to the existence of two states of ES.Ub in an equilibrium ratio 20:l favoring the species with which AMP binds relative to that with which ATP binds. These proposed conformational changes of the enzyme seem plausible on steric grounds in order to account for transfer of ubiquitin between two separate sites. The observation that AMP binds to E S . U b but not ESH with which ATP reacts is consistent with a conformation for free enzyme similar to the minor ES.Ub conformation.
Finally, it is important to note the remarkably high affinity the enzyme displays for ATP and ubiquitin. An ATP dissociation constant, KL,, of 0.45 p~ is significantly lower than the usual intracellular concentration for ATP of 1-2 mM (27). However, this value for KL1 is consistent with previous reports that ATP must fall to levels considerably less than 10% of its physiological concentration before a large effect on energydependent protein degradation is noted (17,(28)(29)(30)(31). Similarly, the low dissociation constant for ubiquitin, KLa, of 0.58 p~ (5 p g / d ) indicates that the enzyme is probably saturated with regard to this substrate even at high steady state levels of conjugates since the intracellular concentration of total ubiquitin has been estimated as 212 p~ in rabbit reticulocytes and human erythrocytes (14).
The present studies have substantiated an earlier minimal mechanism for the ubiquitin activating enzyme (13) and have demonstrated an expanded scheme to be a strictly ordered process. A role for regulation of protein degradation at the level of ubiquitin activation has been proposed involving product inhibition by AMP and PPi. Future studies with purified components of this degradative pathway should allow one to test this hypothesis.
Also, further thermodynamic binding studies should be valuable in assessing the importance of coupling ubiquitin adenylate binding energy to catalysis by this enzyme.

APPENDIX
For Scheme I rewritten as ATP Ub PP