3-Hydroxy-3-methylglutaryl-CoA Synthase PARTICIPATION OF ACETYLS-ENZYME AND ENZYME-S-HYDROXYMETHYLGLUTARYL-SCoA INTERMEDIATES IN THE REACTION*

Replacement of 3-hydroxy-3-methylglutaryl-CoA synthase's glutamate 95 with alanine diminishes catalytic activity by over 5 orders of magnitude. The structural integrity of E95A enzyme is suggested by the observation that this protein contains a full complement of acyl-CoA binding sites, as indicated by binding studies using a spin-labeled acyl-CoA. Active site integrity is also demonstrated by (13)C NMR studies, which indicate that E95A forms an acetyl-S-enzyme reaction intermediate with the same distinctive spectroscopic characteristics measured using wild type enzyme. The initial reaction steps are not disrupted in E95A, which exhibits normal levels of Michaelis complex and acetyl-S-enzyme intermediate. Likewise, E95A is not impaired in catalysis of the terminal reaction step, as indicated by efficient catalysis of a hydrolysis partial reaction. Single turnover experiments indicate defective C-C bond formation. The mechanism-based inhibitor, 3-chloropropionyl-CoA, efficiently alkylates E95A. This is compatible with the presence of a functional general base, raising the possibility that Glu(95) functions as a general acid. Demonstration of a significant upfield shift for the methyl protons of HMG-CoA synthase's acetyl-S-enzyme reaction intermediate suggests a hydrophobic active site environment that could elevate the pK(a) of Glu(95) as required to support its function as a general acid.

laboratory (l-3) have shown that homogeneous avian liver HMG-CoA synthase is acetylated by acetyl-CoA at a cysteinyl-SH giving rise to acetylS-enzyme. This observation and other evidence (4-7) support the proposal that acetylS-enzyme is an intermediate in HMG-CoA synthesis from acetyl-CoA and acetoacetyl-CoA. Based on the premise that acetylation of the enzyme by acetyl-CoA occurs first in the reaction sequence, it follows that condensation with acetoacetyl-CoA, the second substrate in the reaction, would generate the novel intermediate, enzyme&-HMG-SCoA.
A final hydrolytic step would then be required to release HMG-CoA from its thioester linkage to the enzyme. The three-step reaction sequence outlined below (Reactions 2 to 4) accounts for all pertinent experimental facts in the literature bearing on the mechanism of action of HMG-CoA synthase. While the participation of an enzymes-HMG-SCoA intermediate in the reaction had been suggested earlier (2, 41, attempts to demonstrate its existence were unsuccessful. In view of more recent findings (31, including those reported in this paper, it appears that earlier failures to detect this intermediate can be attributed to the slow rate of its formation (Reaction 3) relative to hydrolysis (Reaction 4). In the present investigation, conditions were employed which made it possible to trap and, therefore, to characterize the elusive enzyme-S-HMG-SCoA intermediate. N-(3-Hydroxy-3-methylglutaryl) cysteic acid (N-HMG-cysteic acid) was prepared by mixing cysteine at pH 8 with a l.l-fold molar excess of HMG anhydride prepared as described by Goldfarb and Pitot (14). Acylation was allowed to proceed at 25" for 30 min with constant stirring.
The reaction mix was stored at room temperature overnight to allow completion of the S to N intramolecular rearrangement which occurs spontaneously at neutral pH (15, 16). The sample was then taken to dryness and was oxidized with performic acid (17). After completion of the oxidation, a large excess of water was added and the sample was flash-evaporated to remove formic acid. The overall yield of N-HMG-cysteic acid was 85%. Thin layer chromatography of the concentrated sample, employing a variety of systems (cf. Fig. l Acetyl-S-enzyme Used for Production ofEnzyme-S-HMG-SCoA -Sixty nanomoles of acetyl-CoA or 80 nmol of [l-14Clacetyl-CoA (6500 cpmlnmol) were added to 1.5 nmol of HMG-CoA synthase in a solution (total volume, 0.2 ml) containing 2 pmol of potassium phosphate, pH 7.5, and glycerol (24%, v/v). After incubation at 25" for 5 min to ensure stoichiometric acetylation of enzyme (3), the reaction mixture was quickly cooled to 0" before being frozen at -60" overnight.
Each mixture was subsequently inserted in a -25" bath and 40 ~1 of cold 1 M acetate buffer, pH 5.0, was added, followed by 80 ~1 of cold ethanol bringing the mix (final volume, 0.32 ml) to 25% (v/v) in ethanol and permitting thawing at the reduced temperature.
Reaction with acetoacetyl-CoA was initiated as soon as complete mixing of the sample was achieved. acetyl-S-enzyme, the amount of labeled enzyme-S-HMG-SCoA formed was determined as proteinbound radioactivity by the trichloroacetic acid precipitation and filtration procedure described below. When [l-'4Clacetyl-S-enzyme was reacted with unlabeled acetoacetyl-CoA, it was necessary to eliminate radioactivity due to unreacted ll-14Clacetyl-S-enzyme before a measurement of enzyme-S-l'*C]HMG-SCoA could be made. Condensation of acetylS-enzyme with acetoacetyl-CoA was stopped by addition of cold 10% trichloroacetic acid to the reaction mixtures. Fifty micrograms of bovine serum albumin, which acted as carrier, was then added to each sample. Precipitated protein was centrifuged and then resuspended in cold 10% trichloroacetic acid. The suspension was loaded onto a 2.5-cm diameter glass fiber filter which was then washed 10 times with 5 ml of cold 10% trichloroacetic acid, 6 times with 5 ml of 50 mM sodium pyrophosphate in 0.5 M HCl, and once with 10 ml of cold ethanol. Filters containing 114Clacetyl-labeled protein were oxidized overnight over performic acid vapor to cleave thioester linkages, converting all 14C activity to either 11-"Clacetic acid or [14C]HMG.
Samples were then stored in uacuo over alkali pellets to remove [l-"'Clacetic acid from the filter, permitting accurate quantitation of nonvolatile 14C activity as lZ4ClHMG due to enzyme-S-[Y!]HMG-SCoA.
Filters were then counted in a toluene/ Triton X-loo-containing scintillator fluid, as previously described (2).

Formation
of Enzyme-S-HMG-SCoA from Acetyl-S-enzyme and Acetoacetyl-CoA -It is possible to trap an intermediate having the properties of the condensation product of [*4C]acetyl-S-enzyme and acetoacetyl-CoA by controlling conditions (3) that affect the relative rates of the partial reactions of HMG-CoA synthesis (Reactions 2 to 4). 114C1Acetyl-S-enzyme is prepared by incubating [l-'4Clacetyl-CoA with HMG-CoA synthase for 5 min at 25" and then freezing the mixture at -60". The acetylated form of the synthase is precipitable with trichloroacetic acid and contains approximately one acetyl group per molecule of enzyme (Table IA). Previous investigations (2) showed that the acetyl group of acetyl-enzyme is covalently attached to the enzyme in thioester linkage to a cysteinyl-SH. This acetyl group is cleaved and volatilized upon exposure of acetylS-enzyme to performic acid vapor (Table  IA). When incubated with acetoacetyl-CoA for 10 s at -25" (in the presence of 25% ethanol and 15% glycerol,2 ['4C1acetyl-Senzyme is converted in 16 to 20% yield to a trichloroacetic acid-    A reaction mixture containing 9 nmol of homogeneous HMG-CoA synthase and 47 nmol of [l-"Clacetyl-CoA (6500 cpm/nmol) in 1.2 ml of 8 rnM potassium phosphate buffer, pH 7.5, was incubated for 5 min at 25", then frozen overnight at -60". The reaction mixture was then inserted in a -25" bath and 0.24 ml of cold 1 M acetate buffer, pH 5.0, and 0.48 ml of cold ethanol was added, permitting thawing at the reduced temperature.
Condensation was initiated with 120 nmol of acetoacetyl-CoA and allowed to proceed at -25" for 10 s. Enzyme-S-["CIHMG-SCoA was precipitated, trapped on a glass fiber filter, and was oxidized with performic acid as described under "Experimental Procedures." The oxidized sample on the filter was extracted with cold (0" acetoacetyl-enzyme in the biosynthesis of mevalonic acid. While it is possible that such an adduct forms, but is very labile, our negative findings are in agreement with those of Middleton and Tubbs (7), who, using an indirect acylation assay with the yeast synthase, also failed to detect such an acetoacetyl-enzyme adduct.
Isolation and Characterization of the Active Site Fragment Containing the 3-Hydroxy-3-methylglutaryl Condensation Product-It has been established (1, 2) that the site of acetylation of HMG-CoA synthase by acetyl-CoA is a cysteinyl-SH group. If, as we propose (2, 3), acetyl-S-enzyme reacts with acetoacetyl-Cob in a second step (Reaction 31, the immediate condensation product, i.e. HMG-CoA, should remain covalently bound to the same cysteinyl-SH group of the synthase. This postulate is consistent with the finding that performic acid cleaves the labeled condensation product from the enzyme (Table IA). In order to determine directly whether a cysteinyl-SH is the site of attachment of HMG-CoA, experiments were undertaken to isolate a fragment containing both cysteine and the labeled 3-hydroxy-3-methylglutaryl moiety. The procedure employed is outlined in Scheme 1; the results are summarized in Table III   Reaction mixtures (total volume, 2.0 ml) were quickly cooled to 0" and then frozen overnight at -60". The samples were thawed rapidly at -25" after addition of cold acetate buffer, pH 5.0, (final concentration 125 mM) and then addition of cold ethanol (final concentration 25%, v/v). Final sample volume after these additions was 3.2 ml. The protein in two samples was precipitated by addition of cold 10% trichloroacetic acid. In the third sample, the condensation reaction was initiated rapidly by the addition of 200 nmol of acetoacetyl-CoA and was allowed to proceed for 10 s before termination by the addition of cold 10% trichloroacetic acid. Precipitated protein was trapped on glass fiber filters and washed as described under "Experimental Procedures." One of the two filters containing [l-'4Clacetyl-enzyme was counted directly and the other was used to determine performic acid-stable 14C activity as described under "Experimental Procedures." The filter containing the 14C-labeled condensation product was mechanically disrupted in 2 ml of 20 mM imidazole chloride buffer, pH 6.8, containing 1 mg of pronase and 5% ethanol. Proteolysis was allowed to proceed for 69 h at 37" after which the digest was centrifuged and the pellet washed with imidazole buffer. The pooled supernatant solution and wash, containing essentially all of the 14C activity initially trapped, was lyophilized, dissolved in concentrated formic acid containing 1.5% hydrogen peroxide, and incubated overnight at 4". The sample was then taken to dryness, the residue dissolved in HCl, pH 1.5, and loaded on a Dowex 5OW-X4 column (12 x 0.9 cm) previously equilibrated with an HCl solution at pH 1.5. The "'C activity was eluted as a single peak upon washing the column with HCl, pH 1.5. The 14C-containing fractions were taken to dryness and subjected to thin layer chromatographic characterization (Fig. 1). AcAc-refers to acetoacetyl-and TCA to trichloroacetic acid.

Acyl-enzyme
Intermediates in HMG-CoA Synthesis N-acetylcysteamine (16) and of S-acetylcysteine to N-acetylcysteine (2). This rearrangement explains the finding that proteolysis at neutral pH renders all of the 14C-labeled acyl groups stable to per-formic acid treatment. To aid in the subsequent isolation and characterization of the labeled acylated residues, the entire pronase digest was subjected to performic acid oxidation. After lyophilization of this mixture, the residue was redissolved, brought to pH 1.5, and passed through a column of Dowex 50 (H+ form) to remove free amino acid contaminants.
Nearly all (87%) of the 14C applied to the column was recovered indicating that the labeled derivative(s) in the eluate did not possess free amino groups. Using several thin layer chromatographic systems (Fig. l), co-chromatography of the concentrated eluate with authentic N-acetylcysteic acid and N-(3-hydroxy-3-methylglutaryl) cysteic acid reveals two major radioactive peaks. Based upon the coincidence of the two major radioactive peaks with those of the N-acylcysteic acid standards, it is concluded that the 14C-labeled derivatives are N-acetyl-and N-(3-hydroxy-3-methylglutaryl) cysteic acid. N-HMG-cysteic acid constitutes approximately 27% of the total radioactivity in the sample. This observation is in good agreement with the 25% estimated above by direct analy- ['4Clacyl-enzyme precipitates would have been readily distinguished by thin layer chromatography. All data generated in this series of experiments are consistent with the proposal that the condensation product of the HMG-CoA synthase-catalyzed reaction is attached to the enzyme at an active site cysteine.
Relative Rates of Acetyl Transfer and Condensation Reactions of Acetyl-S-enzyme -The reactions in which acetylSenzyme may participate when incubated in the presence of CoA and acetoacetyl-CoA are indicated in Scheme 2. Knowledge of the rate of condensation of acetyl-S-enzyme with acetoacetyl-CoA, relative to the rate of acetyl transfer from acetyl-S-enzyme to CoA, would provide an independent means of evaluating the role of acetyls-enzyme in HMG-CoA synthesis (Reaction 1). It has already been established (2) that the rate of acetyl transfer (km,, Scheme 2) is at least as fast as the rate of HMG-CoA formation from acetyl-CoA and acetoacetyl-CoA. Thus, if condensation (k2, Scheme 2) were as rapid as acetyl transfer, acetyl-S-enzyme would meet the kinetic requirements of a participant in the overall process. The relative rates of the competing condensation and acetyl transfer reactions were estimated by using a variation of the isotope trapping technique developed in Meister's laboratory (22,23). To monitor the relative rate of the alternative reactions of the proposed intermediate, ['4Clacetyl-S-enzyme is mixed simultaneously with CoA, a substoichiometric amount of acetoacetyl-CoA, and a large excess of unlabeled acetyl-CoA. Upon mixing, any ['4Clacetyl group transferred to CoA will undergo extensive isotope dilution with the unlabeled acetyl-CoA pool. Hence, acetylS-enzyme formed by reacetylation will have a greatly reduced specific activity. Alternatively, [W]acetyl-S-enzyme may initially condense with acetoacetyl-CoA to form a labeled product with the same specific activity as the starting [14C]acetyl-S-enzyme. By maintaining acetoacetyl-CoA at levels that are substoichiometric with respect to that of the ['4Clacetyl-S-enzyme added, multiple productive enzyme turnovers cannot occur since the condensation reaction is virtually irreversible (4, 24). Hence, there is no need to rapidly quench the reaction to prevent multiple turnovers which would complicate interpretation of the results. A comparison of the specific activity of the [14C]HMG-CoA formed (including enzyme-S- ['*ClHMG-SCoA) with that of the [14Clacetyl-S-enzyme initially present provides a measure of the extent of isotope dilution of ['4Clacetyl-S-enzyme (by unlabeled acetyl-CoA) prior to condensation. The extent of isotope dilution is a measure of the rate of acetyl transfer relative to condensation. If the condensation process is very rapid, the specific activity of ['QHMG-CoA will have a specific activity comparable to that of the isotopediluted ['Qacetyl-CoA pool. Fig. 2 illustrates the kinetics and CoA concentration dependence of acetyl transfer from acetylS-enzyme to CoA in the absence of acetoacetyl-CoA.
It is evident that when the CoA concentration is greater than 1.0 mM, acetyl transfer reaches completion in less than 10 s. In all subsequent isotope trapping experiments, the CoA concentration was maintained at 1.6 mM to ensure quantitative acetyl group transfer from CoA concentration dependence of the rate of "C-labeled acetyl transfer from ll-'"Clacetyl-S-enzyme to CoA. HMG-CoA synthase (0.7 nmol) was mixed with l'4Clacetyl-CoA (7 nmol, 6500 cpm/ nmol) in 100 mM potassium phosphate, pH 7.0, resulting in the formation of [14Clacetyl-S-enzyme (0.7 nmol, 6500 cpmlnmol). '%labeled acetyl transfer was initiated by adding a 20-fold excess (140 nmol) of unlabeled acetyl-CoA, together with varying amounts of CoA to produce the final concentrations indicated in the figure (final volume, 0.09 ml). InA, W-labeled acetyl transfer was terminated by addition of trichloroacetic acid after the reaction had proceeded for 10 s at 30". In B, 14C-labeled acetyl transfer was allowed to proceed for the times indicated before termination of the reaction by addition of trichloroacetic acid. The amount of '%-activity remaining as ['4C]acetyl-S-enzyme was determined by filtering the precipitated enzyme on glass fiber filters and counting the filters as described under "Experimental Procedures." before condensation (i.e. were K-, >> k, in Scheme 2). The points on Line B (Fig. 3) were obtained in an experiment in which both the condensation and acetyl transfer routes were available to the added [14C]acetyl-S-enzyme, i.e. both acetoacetyl-CoA and unlabeled acetyl-CoA were present. While the ratio of any pair of points on Lines A and B obtained at the same concentration of acetoacetyl-Cob may be used to determine the relative rates of condensation and acetyl transfer (12,and Km,,respectively,in Scheme 2), the averaging achieved by comparing the slopes of lines fit to these points provides a better measure of the relative rates. The slope of Line B is twothirds that of Line A, indicating that, under the conditions of this experiment, condensation (k2, Scheme 2) is twice as rapid as acetyl transfer (K-,, Scheme 2). In view of our earlier finding (2, 6) that K-, (Scheme 2) is as rapid as the overall process, i.e. the formation of HMG-CoA from acetyl-CoA and acetoacetyl-CoA, it follows that condensation (k,) is at least as rapid as the overall reaction. It can be concluded, therefore, that acetylS-enzyme meets the kinetic criteria for participation as an intermediate in HMG-CoA synthesis.

DISCUSSION
The finding that HMG-CoA synthase is rapidly acetylated by acetyl-CoA at a cysteinyl-SH ( 1, 2)  sation and hydrolysis steps has a sound chemical basis. Prior hydrolysis of the acetyl-enzyme thioester linkage would cause premature loss of the carbanionic character of the acetyl methyl group necessary for condensation. Moreover, it would lead to the loss of the high acyl transfer potential needed to pull the thermodynamically unfavorable condensation. The failure of earlier attempts (4) to detect enzyme-S-HMG-SCoA suggests a rate of hydrolysis (Reaction 4) which is rapid relative to the rate of condensation (Reaction 3). While investigating the rates of the partial reactions, it became evident (3) that the avian liver HMG-CoA synthase undergoes a reversible temperature-dependent conformational change which affects the rate of first partial reaction, i.e. acetylation of the enzyme (Reaction 2). It was established that a conformational change in the enzyme affected both its capacity for rapid acetylation by acetyl-CoA and ability to catalyze the synthesis of HMG-CoA. The possibility was considered that the relative rates of the other partial reactions, i.e. condensation and hydrolysis, might also be affected by altering the thermal history of the enzyme and the temperature at which the condensation reaction is performed. Preliminary studies (3, 8) verified the feasibility of this approach to the trapping of the postulated enzyme-S-HMG-SCoA intermediate. Therefore, in the present investigation, the condensation of acetyl-s-enzyme with acetoacetyl-CoA was performed under conditions which block the first partial reaction (Reaction 2) and increase the ratio of the rates of condensation (Reaction 3) to hydrolysis (Reaction 4). From the results of these experiments (Table III), it can be calculated that condensation exceeded hydrolysis by a factor of nearly 2. Of the 94,000 cpm of ['4C]acetyl-S-enzyme present initially, 21,000 cpm (i.e. 94,000 -73,000 cpm) were apparently lost via hydrolysis of enzyme-S-[14C]HMG-SCoA (Reaction 4) while 18,000 cpm were recovered as enzyme-S-['4C1HMG-SCoA in the per-formic acid-treated enzyme precipitate. Therefore, the total amount of enzyme-S-["'CIHMG-SCoA formed during condensation was 39,000 cpm (i.e. 18,000 + 21,000 cpm). Hence, condensation must have occurred at a rate almost twice that of hydrolysis. These results, together with those previously reported (3), support the hypothesis that the condensation/hydrolysis ratio can be increased through alterations in the conditions under which the condensation reaction is performed.
Additional information concerning the relative rates of the partial reactions was obtained from isotope trapping experiments (Fig. 3) which indicate that the condensation of acetyl-S-enzyme with acetoacetyl-CoA is more rapid than acetyl transfer from acetyl-S-enzyme to CoA (Scheme 2). If previous estimates of&, = 1 for the acetylation reaction are correct (2, 7), it may be inferred that condensation occurs more rapidly than acetylation. While direct measurements of the rate of hydrolysis of enzyme-S-HMG-SCoA have not been made, the difficulties encountered in trapping appreciable amounts of this species suggest that hydrolysis is extremely rapid, particularly at 25". Thus, it is likely that formation of acetyl-Senzyme (Reaction 2) is the rate-limiting step in HMG-CoA synthesis. A conformational change may well be involved in converting the synthase to the "active," i.e. acetylatable, form required to initiate a catalytic cycle. Several lines of evidence are consistent with this possibility. First, acetylation of the enzyme (Reaction 2) appears to be the slow step in the overall process. Secondly it has been established (3) that the synthase can be thermally perturbed, i.e. inactivated, by lowering the temperature and that the return to a catalytically active state (or conformation) at a temperature ~25" is exactly correlated with the return of the capacity of the enzyme to undergo acetylation by acetyl-CoA. Direct demonstration that a conformational change occurs prior to acetylation of enzyme and proof that the kinetics of such a change determines the rate of acetylation (and, therefore, the rate of the overall reaction) will require additional experimentation.
Fluorescence or spinlabeling approaches could prove valuable in monitoring changes in protein conformation over a wide range of temperature. Attempts are currently being made to develop an appropriate probe for such studies.