Investigation of Conserved Acidic Residues in 3-Hydroxy-3-methylglutaryl-CoA Lyase

3-Hydroxy-3-methylglutaryl-CoA (HMG-CoA) lyase catalyzes the divalent cation-dependent cleavage of HMG-CoA to form acetyl-CoA and acetoacetate. In metal-dependent aldol and Claisen reactions, acidic residues often function either as cation ligands or as participants in general acid/base catalysis. Site-directed mutagenesis was used to produce conservative substitutions for the conserved acidic residues Glu-37, Asp-42, Glu-72, Asp-204, Glu-279, and Asp-280. HMG-CoA lyase deficiency results from a human mutation that substitutes lysine for glutamate 279. The E279K mutation has also been engineered; expression in Escherichia coli produces an unstable protein. Substitution of alanine for glutamate 279 produces a protein that is sufficiently stable for isolation and retains substantial catalytic activity. However, thermal inactivation experiments demonstrate that E279A is much less stable than wild-type enzyme. HMG-CoA lyase deficiency also results from mutations at aspartate 42. Substitutions that eliminate a carboxyl group at residue 42 perturb cation binding and substantially lower catalytic efficiency (104-105-fold decreases in specific activity for D42A, D42G, or D42H versus wild-type). Substitutions of alanine for the other conserved acidic residues indicate the importance of glutamate 72. E72A exhibits a 200-fold decrease in kcat and >103-fold decrease in kcat/Km. E72A is also characterized by inflation in the Km for activator cation (26-fold for Mg2+; >200-fold for Mn2+). Similar, but less pronounced, effects are measured for the D204A mutant. E72A and D204A mutant proteins both bind stoichiometric amounts of Mn2+, but D204A exhibits only a 2-fold inflation in KD for Mn2+, whereas E72A exhibits a 12-fold inflation in KD (23 μm) in comparison with wild-type enzyme (KD = 1.9 μm). Acidic residues corresponding to HMG-CoA lyase Asp-42 and Glu-72 are conserved in the HMG-CoA lyase protein family, which includes proteins that utilize acetyl-CoA in aldol condensations. These related reactions may require an activator cation that binds to the corresponding acidic residues in this protein family.

HMG-CoA 1 lyase catalyzes the cleavage of HMG-CoA into acetyl-CoA and acetoacetate (1) by a proposed general acid/base mechanism (Scheme 1). Enzyme activity is absolutely dependent on the presence of a divalent cation (e.g. Mg 2ϩ , Mn 2ϩ ) and is stimulated by reducing agents (e.g. DTT). Cleavage of HMG-CoA is involved in the generation of ketone bodies to maintain the energy requirements of non-hepatic tissues (2) as well as the terminal step of leucine catabolism (3).
HMG-CoA lyase is a member of the HMG-CoA lyase family of proteins that catalyze C-C cleavage/condensation reactions. Members of the HMG-CoA lyase family include homocitrate synthase and isopropylmalate synthase; both catalyze the aldol condensation of acetyl-CoA with either ␣-ketoglutarate or ␣-ketoisovalerate, forming homocitrate and ␣-isopropylmalate, respectively (4). A molecular structure has not yet been determined for any family member, and few details regarding the enzyme reaction chemistry have been reported for these family members. Therefore, studies of HMG-CoA lyase may improve our basic understanding of structure/function relationships among members of the HMG-CoA lyase family.
The importance of the ketogenic cycle is underscored in hereditary HMG-CoA lyase deficiency, which can result in hypoketotic hypoglycemia and a marked increase in serum levels of several organic acids. Uncontrolled, HMG-CoA lyase deficiency can result in mental retardation and episodes of seizures and coma (5). Several mutations in the HMG-CoA lyase gene correlating with deficiency have been identified (5), including the missense mutations: H233R (6), R41Q, D42E, D42H, D42G (7), and most recently, E279K (8). Modeling of human mutations using a recombinant expression system would allow for the more detailed characterization of the mutant enzyme, testing the correlation of the mutation with human disease and potentially identifying active site residues.
As noted above, HMG-CoA lyase activity requires the presence of a divalent cation, such as Mg 2ϩ or Mn 2ϩ . Previous experiments have suggested that the divalent cation and substrate form a ternary complex with the enzyme. The metal has been proposed to ligate to oxygen atoms of the 1-carboxyl and 3-hydroxyl group of HMG-CoA, facilitating enolization and thereby, stabilization, of a carbanion intermediate (9). Two or more amino acid side chains are expected to function directly or indirectly in the octahedral coordination of the cation activator, but to date, only one amino acid, histidine 235, has been implicated as an enzyme ligand to the metal (10).
Cysteine 266 and histidine 233 have been identified as catalytic amino acids based on mutagenic and kinetic analyses (6,11). Upon mutation to produce conservative substitutions (C266S and H233A), the enzymes exhibit 10 3 -and 10 4 -fold diminutions in k cat , respectively, without a significant effect on K m for either HMG-CoA or metal. However, whereas substantial diminutions in catalytic efficiency have been demonstrated, precise functional assignments have not yet been made and it is possible that the general base and general acid catalysts, as well as potential additional cation ligand(s), remain to be identified.
Precedent in other Claisen-condensing enzyme reactions, such as malate synthase and citrate synthase, have suggested multiple roles for acidic residues in catalysis (12) or cation ligation (13,14). These observations and other precedents with metal requiring (Class II) aldolases (15) prompted mutagenesis of acidic residues (Glu-37, Asp-42, Glu-72, Asp-204, Glu-279, and Asp-280) in HMG-CoA lyase. These residues are highly conserved in HMG-CoA lyase proteins, as well as many of the proteins in the HMG-CoA lyase family, and/or have been implicated by mutations that correlate with human enzyme deficiency. The results of these experiments address some functional issues concerning residues implicated in human inherited disease and implicate several acidic amino acids as active site residues that support cation ligation or influence reaction efficiency. A preliminary account of part of this study has appeared (16).

EXPERIMENTAL PROCEDURES
Materials-Escherichia coli (JM109) cells were obtained from Promega. High performance liquid chromatography-purified deoxyoligonucleotides were purchased from Operon. Qiagen plasmid kits (Qiagen Inc.) were used to isolate plasmid DNA from bacterial cultures. Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs. Bacto-tryptone and yeast extract are products of Difco Laboratories. Isoproyl-1-thio-␤-D-galactoside was bought from United States Biochemical Corp. Phenyl-agarose was purchased from Sigma and Q-Sepharose anion exchange resin from Amersham Biosciences. Hydroxyapatite Bio-Gel HTP resin was obtained from Bio-Rad. Malate dehydrogenase and citrate synthase were obtained from Sigma. HMG-CoA was synthesized from the anhydride, prepared from the free acid (Fluka) according to Goldfarb and Pitot (17); concentration of (S)-isomer of HMG-CoA was determined by enzymatic end point analysis. 3-Hydroxyglutaryl-CoA was synthesized as described by Kramer and Miziorko (18). [ 14 C]HMG-CoA (8700 dpm/nmol) was enzymatically synthesized from acetoacetyl-CoA and 1-[ 14 C]acetyl-CoA as described by Vinarov and Miziorko (19). All other chemicals were reagent grade.
Site-directed Mutagenesis-The expression construct for HMG-CoA lyase (pTRC-HL1) is derived from the pTRC-99a expression vector, as previously described (20). cDNA for generation of the D42E, D42G, and D42H mutant proteins was kindly provided by Dr. Grant Mitchell (Hopital Ste. Justine-Univ. of Montreal). Generation of the remaining point mutations in HMG-CoA lyase was performed by full-circle PCR using the QuikChange TM site-directed mutageneis kit from Stratagene. The following primers were used to introduce the site-specific mutations by full-circle PCR: E37A, 5Ј-GGTGAAAATTGTGGCGGTTGGTC-CCCGAG-3Ј; E37D, 5Ј-GGTGAAAATTGTGGATGTTGGTCCCCGAG-3Ј; D42A, 5Ј-GGAAGTTGGTCCCCGAGCGGGACTACAAAATG-3Ј; D42N, 5Ј-GGAAGTTGGTCCCCGAAACGGACTACAAAATG-3Ј; E72A, 5Ј-CTCTCTGTTATAGCAACCACCAGCCTTG-3Ј; D204A, 5Ј-ATCTCC-CTGGGGGCCACCATTGGTGTG-3Ј; E279A, 5Ј-GAAACTTGGCCACA-GCAGACCTGGTCTAC-3Ј; E279K, 5Ј-GAAACTTGGCCACAAAAGAC-CTGGTCTAC-3Ј; D280A, 5Ј-GGCCACAGAAGCCCTGGTCTACATGC-3Ј. Only the 5Ј to 3Ј directional primers have been listed above. The boldface and underlined characters indicate the base changes that were made to generate mutagenic codons. Mutagenic plasmid DNA was isolated from selected transformants and analyzed by restriction mapping and DNA sequencing. To ensure no secondary mutations were generated during the PCR, either the complete gene was sequenced or a restriction fragment containing the desired point mutant was sequenced and ligated into the plasmid to replace the corresponding restriction fragment of the wild-type gene. DNA sequencing was performed by the Protein and Nucleic Acid Facility at the Medical College of Wisconsin. The verified plasmids were transformed into competent JM109 cells for subsequent expression of mutant lyase enzymes.
Expression and Purification of Wild-type and Mutant Enzymes-Enzymes were expressed and purified from 3-liter cultures as described by Roberts et al. (20) with two modifications. First, induction of bacterial cultures by isoproyl-1-thio-␤-D-galactoside was started at A 600 ϭ 1.0 rather than 0.6. Second, the phenyl-agarose pool for D280A and E279A proteins was applied to a 1 ϫ 5-cm hydroxyapatite column to remove a contaminant that persisted throughout the prior chromatography steps. The hydroxyapatite column was equilibrated with 20 mM potassium phosphate, pH 7.8, containing 20% glycerol and 1 mM DTT. After loading, the column was washed with 10 column volumes of equilibration buffer. HMG-CoA lyase protein was eluted using 8 column volumes of a 20 -200 mM potassium phosphate, pH 7.8, gradient; gradient buffers contained 20% glycerol and 1 mM DTT. Enzyme purity for wild-type and each mutant enzyme was assessed by SDS-PAGE and Coomassie Blue staining. SDS-PAGE was performed according to Laemmli (21) using an 11.5% acrylamide running gel and a 4.5% acrylamide stacking gel.
Protein and Enzyme Assays for HMG-CoA Lyase-Protein concentration was determined following the method of Bradford (22) using the Bio-Rad protein assay reagent (Bio-Rad) and bovine serum albumin as the standard. A subunit molecular mass of 33,000 Da was used to calculate the molar concentration of enzyme. Modification of enzyme with [ 14 C]butynoyl-CoA (16,000 dpm/nmol) was performed as previously described by Hruz and Miziorko (23).
HMG-CoA lyase enzymatic activity was measured using the citrate synthase-coupled assay of Stegink and Coon (1) as modified by Kramer and Miziorko (24). All reaction components, including buffer, water, and substrate were passed over a Chelex 100 column (Bio-Rad) to remove trace metals. When kinetic constants for Mn 2ϩ were evaluated, DTT was omitted from the assay mixture to prevent interaction of metal with the reductant. In this case, enzyme was incubated separately from the assay mixture in buffer containing 5 mM DTT and added 2 min before the start of the reaction assay. Specific activity was determined for D42A lyase using a radioactive assay (25) to improve sensitivity. V max , k cat , and K m values for HMG-CoA lyase were determined from initial velocity data that were fit by non-linear regression analysis using the Grafit program (Erithacus software).
ESR Spectroscopy-The binding of Mn 2ϩ to human HMG-CoA lyase was measured on a Varian Century Line 9 GHz spectrometer with a TE 102 cavity as described by Roberts and Miziorko (10). Each spectrum was recorded at 22°C with a modulation amplitude of 10 G, a modulation frequency of 100 kHz, a microwave power of 60 mW, a field sweep of 1000 G, and a time constant of 0.25 s. Prior to ESR, protein samples were concentrated to 200 M and glycerol was removed using a Sephadex G-25 centrifugal column equilibrated with 25 mM Tris-Cl, pH 8.2, containing 150 mM NaCl and 1 mM DTT. A quartz capillary tube was used for all measurements. The ESR samples contained variable concentrations of Mn 2ϩ (10 -150 M) with constant HMG-CoA lyase sites. The amount of bound Mn 2ϩ was determined by directly comparing the spectral amplitudes of samples containing HMG-CoA lyase to the corresponding amplitudes observed with a buffered solution containing an equal concentration of Mn 2ϩ in the absence of enzyme. Scatchard plots of the data were subjected to linear regression analysis using Grafit to determine both the Mn 2ϩ binding stoichiometry and the binding affinity of the HMG-CoA lyase mutants.

Strategy for Identification of Acidic Amino Acids as Mutagenesis Targets-
The reports (7,8) that point mutations at aspartate 42 and glutamate 279 correlate with HMG-CoA lyase deficiency prompted mutagenesis tests of these residues as well as the flanking residue, aspartate 280. In the HMG-CoA lyase family of enzymes (HMGL-like PF00682), which catalyze related C-C bond condensation/cleavage reactions, alignment of several HMG-CoA lyases with HMGL-like family members allowed the identification of highly conserved acidic residues that include not only aspartate 42, but also glutamate 37, glutamate SCHEME 1. Reaction catalyzed by HMG-CoA lyase.
72, and aspartate 204 (Fig. 1). This alignment was generated using the Pfam data bank (26). Therefore, glutamate 37, glutamate 72, and aspartate 204 have also been investigated by mutagenesis approaches. The human mutations D42E, D42G, D42H, and E279K were modeled by producing recombinant forms of these proteins; other substitutions were produced as necessary to evaluate the importance of residues 42 and 279 to enzyme function. In the case of other conserved acidic residues, alanine substitution was employed.
Expression and Isolation of Mutant HMG-CoA Lyase Proteins-With the exception of E279K and E37A, the mutant enzymes were expressed and purified using the protocol developed for wild-type mitochondrial HMG-CoA lyase (20). Specific activities of purified proteins are documented in Table I. As indicated in Fig. 2, the mutant enzymes that were subjected to detailed characterization were isolated in a highly purified form. In the case of either E279K or E37A, Western blot analysis detected protein in bacterial extracts but after subjecting material to the initial purification steps these mutant proteins were no longer detectable, presumably because of instability. This prompted construction, expression, and isolation of the more conservative E279A and E37D substitutions. Alanine substitution for either glutamate 279 or the adjacent aspartate 280 as well as aspartate substitution for glutamate 37, produces enzymes that are sufficiently stable to permit isolation and measurement of specific activities ( Table I) that are comparable with estimates for wild-type enzyme. However, incubation of purified E279A at 37°C (Fig. 3) resulted in a significant decay in measurable activity (t1 ⁄2 ϭ 60 min), whereas comparable treatment of wild-type enzyme did not result in loss of activity. In fact, wild-type enzyme retained full activity after a 6-h incubation. Other HMG-CoA lyase mutants (e.g. E72A, D204A, and D280A) exhibit stability comparable with wild-type enzyme. On the basis of the results with E279K and E279A, the HMG-CoA lyase deficiency associated with mutation at this residue can be attributed to impaired stability of the enzyme.
Other reports of human HMG-CoA lyase deficiency (7) implicate highly conserved aspartate 42 as an important residue. In inherited human HMG-CoA lyase deficiency, the point mutations D42E, D42G, and D42H have been identified. As indicated below, a variety of substitutions (including the clinically detected mutations) can be engineered for this aspartate and the resulting mutant proteins remain sufficiently stable for characterization in some detail. As indicated in Table I, specific activity of any particular aspartate 42 mutant is strongly dependent on the nature of the substitution; any replacement of the acidic side chain results in a marked decrease in specific activity.
Substantial decreases in specific activity are also observed ( Table I) upon substitutions that replace the acidic side chain of either glutamate 72 or aspartate 204, both of which are highly conserved in the HMG-CoA lyase family of enzymes. To test whether those mutants that exhibit large decreases in catalytic activity (e.g. D42A, D42G, D42H, D42N, E72A, D204A) retain significant structural integrity, modification of these proteins with the affinity label, 2-butynoyl-CoA, was performed. The modification stoichiometry for these low activity mutants is not substantially different from the value measured for wild-type enzyme ( Table I), suggesting that the active site of each of the mutants approximates the structure of wild-type enzyme and that these mutants are not significantly structurally perturbed. Thus, substantial differences in kinetic or equilibrium binding characteristics that these mutant enzymes exhibit can be interpreted without the concern that a major structural pertur-   (Table II) of the kinetic properties of E37D, E279A, and D280A mutants suggested little difference in V max or K m,HMG-CoA values in comparison with wild-type enzyme. For the clinically significant D42E mutant, there is little change in the apparent K m,HMG-CoA , but almost an order of magnitude decrease in V max . These observations resolve the question raised earlier (7) concerning whether changes in catalytic efficiency or substrate affinity accounted for low enzyme activity in HMG-CoA lyase deficiency that correlates with this particular substitution. The large (4 -5 orders of magnitude) decrease in specific activity (Table I) observed for the D42A, D42G, and D42H substitutions precluded detailed analysis under suboptimal substrate concentrations. Therefore, a D42N mutant, representing a more conservative substitution that is hydrophilic in nature and retains hydrogen bonding ability, was generated to test the importance of this residue. Analysis of D42N (as well as D42E) indicated little change in apparent K m,HMG-CoA (Table II) but a 600-fold decrease in V max when the amide side chain of asparagine replaces the aspartate side chain. Although the effect is smaller in magnitude than the decrease in activity measured for D42A, D42G, or D42H, the substantial effect observed for the conservative asparagine substitution underscores the importance of a carboxyl side chain.
Of those mutant HMG-CoA lyase enzymes that exhibit notable decreases in catalytic efficiency (and for which K m can be accurately estimated), 2 only E72A is characterized by a substantial (1 order of magnitude) inflation in apparent K m,HMG-CoA . To test whether the K m perturbation reflected a change in substrate affinity, the K i for the competitive inhibitor, 3-hydroxyglutaryl-CoA, was measured for this mutant (Fig. 4). This K i , which should closely correlate with the intrinsic affinity of the inhibitor, is inflated for E72A (Table II) by approximately 1 order of magnitude (in good agreement with the magnitude of the K m effect) suggesting an active site function for the carboxyl side chain of glutamate 72.
Analysis of kinetic and equilibrium binding parameters has been performed to evaluate the interaction of divalent cation with the catalytically impaired mutants (Table III). D204A exhibits inflation of the apparent K m values for Mg 2ϩ and Mn 2ϩ of ϳ20and 21-fold, respectively. Inflations in apparent K m values are observed for D42E (25-and 17-fold) and D42N (80and 19-fold) mutants. E72A also exhibits somewhat larger effects. For the E72A mutant, increases in the apparent K m values for Mg 2ϩ and Mn 2ϩ are ϳ26and 237-fold, respectively.
The combination of catalytic and binding terms in K m estimates can complicate comparisons between wild-type and mutant enzymes, especially when mutants exhibit large changes in catalytic efficiency. A more direct evaluation of the ability of wild-type or mutant HMG-CoA lyases to specifically form a binary enzyme-metal complex is afforded using electron spin resonance to measure Mn 2ϩ binding to enzyme. Results of such experiments were subjected to Scatchard analyses (Fig. 5). D204A retains the ability to form stoichiometric levels of a binary enzyme-metal complex (Fig. 5A) with a K D (3.5 M) that represents only a 2-fold increase over the value measured for wild-type enzyme (Table III) and is consistent with the apparent K m for Mn 2ϩ . In contrast, E72A forms stoichiometric levels of a binary enzyme-metal complex (Fig. 5B) but exhibits Ͼ12fold inflation in the measured K D (23 M; Table III). Inclusion of the substrate analog hydroxyglutaryl-CoA does not substantially change the binding stoichiometry or affinity measured for E72A. In contrast, Mn 2ϩ binding studies on D42N fail to detect a binary enzyme-metal complex; only when hydroxyglutaryl-CoA (1 mM) is included does Mn 2ϩ bind at stoichiometric levels ( Table III). The D42G mutation associated with human inherited disease also affects cation binding; no binary enzymemetal complex is detectable. As in the case of D42N, binding studies performed in the presence of hydroxyglutaryl-CoA (1 mM) detected stoichiometric binding of cation to D42G but the measured K D (27 M) is substantially larger than when such an experiment is performed for D42N (K D ϭ 3.2 M). The more conservatively substituted D42E mutant forms a binary enzyme-Mn 2ϩ complex with stoichiometry and affinity values that are comparable with wild-type enzyme (Table III). DISCUSSION The correlation of metabolic disease or dysfunction with mutations in HMG-CoA lyase that reduce enzyme activity by 10-fold or more has suggested that substantial activity is required for humans to function well under metabolic stress that leads to increased lipid catabolism. In the case of the E279K mutation, transient expression studies (8) suggested a residual level of activity of ϳ2%, but no data were reported on enyzme protein levels or stability. Our results on the bacterial expression of E279K and E279A mutants suggest that altered stability, rather than involvement of Glu-279 in reaction chemistry accounts for lack of activity in these mutants. Thus, this region of the protein may be structurally sensitive, but there is no indication that it contributes to catalytic or regulatory sites of the enzyme. In contrast, mutations at aspartate 42 do not seem to substantially destabilize protein but instead influence the efficiency of the reaction. The random human mutations that correspond to D42H and D42G proteins do not destabilize protein but do correlate with extremely low (Ͻ0.01%) activity. The human D42E mutation does not impede production of stable purified protein with substantial (10%) activity (Tables I and  II) under optimized assay conditions. Although no large change in substrate K m is apparent, a significant increase in K m for the 2 Low residual activity precluded precise K m measurements for D42G, D42H, and D42A but approximate estimates for D42G and D42A did not suggest any large (Ͼ3-fold) effects on K m,HMG-CoA . activator cation (Table III) together with intrinsically lower catalytic efficiency might account for the decreased activity of D42E in a physiological context. This possibility is supported by the observed failure of D42N and D42G to bind activator cation in the absence of a substrate/substrate analog and may account for the observation of low activity in the other clinically detected Asp-42 mutants. Limited characterization of the D42G mutant has determined a K D,Mn of 27 M (14-fold inflation versus wild-type). However, as with D42N, this required the inclusion of hydroxyglutaryl-CoA (1 mM) to detect metal binding to enzyme.
In contrast to the results for glutamate 279 and aspartate 280, which suggest no substantial catalytic or regulatory function for these residues, the data for mutations that eliminate other highly conserved acidic side chains suggest that some of these residues (most notably aspartate 42 and glutamate 72) map within the catalytic site. The inflation in the K D E72A for the binary complex with cation suggests that the carboxyl side chain is one ligand to the cation. The K D for cation binding is not altered in the presence of a substrate analog. However, the affinity for substrate or substrate analog, hydroxyglutaryl-CoA, is weakened. Together, the data suggest that the cation activator bridges the carboxyl side chain and the substrate. Because this carboxyl side chain is conserved in the various enzymes included in the HMG-CoA lyase family, the corresponding acidic residues may support similar functions. Typically, divalent cations such as Mg 2ϩ or Mn 2ϩ exhibit octahedral coordination, with two or more amino acid side chains ligating metal directly or indirectly (e.g. with a water molecule interposed). Therefore, although glutamate 72 has been implicated in such a function, other candidates need to be identified. D204A exhibits inflated K m values for cation (measured kinetically under turnover conditions) but this mutant exhibits equilibrium binding of Mn 2ϩ that is not much different from observations made with wild-type enzyme. Thus, functional assignment of aspartate 204 remains somewhat ambiguous.
In contrast, an active site role for aspartate 42 seems more apparent. Whereas glutamate substitution for aspartate does not have a crucial effect on activity or formation of a binary enzyme-cation complex, substitutions that eliminate a charged carboxylate side chain dramatically lower catalytic efficiency. Precedent suggests that magnitude of the effect on catalysis attributable to elimination of a cation ligand can range from 1 to 4 orders of magnitude. In the case of phosphoenolpyruvate mutase (27), alanine substitution of any of three different acidic residues that function as cation ligands decreases activity by 200-, 500-, and 8900-fold, depending on which acidic function is eliminated. For acetohydroxyacid synthase (28), substitution of asparagine for an aspartate that ligands to the cation required for the reaction decreases activity by only 7-fold. A similar asparagine substitution for either of the two different aspartates that ligand to the metal bound in xylose isomerase (29) lowers activity by 20 -30-fold. Because of such a range of effects on catalysis, the impact of mutagenesis on K m values for cation can be variable and direct equilibrium binding measurements become useful. In the case of D42N and D42G, replacement of the acidic side chain abolishes the ability to detect a binary enzyme-Mn 2ϩ complex. Only in the presence of substrate analog can Mn 2ϩ binding be detected. This observation is analogous to results for histidine 235 (10). Elimination of this imidazole side chain disrupted binary enzyme-cation complex formation; a weak complex could be measured in the presence of substrate analog. Previous results (9) have suggested that both HMG-CoA lyase and its substrate may provide ligands to cation. It seems possible that both aspartate 42 and histidine 235 are involved directly or indirectly in cation ligation, which becomes impaired when either of these side chains is replaced. In the presence of the substrate/substrate analog, additional cation ligands are present (e.g. the C3 hydroxyl or C1 thioester carbonyl groups) and the ability to bind cation improves.
Precedent for cation liganding to carboxyl side chains as well as to substrate in reactions that involve C-C bond formation or cleavage is available for both cation-dependent (class II) aldolases (15) and Claisen condensation reactions (13,14). In particular, results for the analogous malate synthase reaction seem quite relevant. Whereas this reaction involves cation-dependent condensation of acetyl-CoA with glyoxylate to transiently form malyl-CoA that is hydrolyzed to the products malate and CoASH, the condensation is chemically similar to a  reversal of the cation-dependent cleavage of HMG-CoA to form acetyl-CoA and acetoacetate. In recent work that demonstrates the importance of carboxyl groups as cation ligands (14), the importance of cation in polarizing the glyoxylate carbonyl and positioning substrates at the active site has also been indicated. Other work on another form of malate synthase (13) has led to the suggestion that cation may also stabilize the enolate of acetyl-CoA, although this issue remains somewhat speculative. Although the malate synthase results argue that postulating a role for the activator cation of HMG-CoA lyase in ligating both carboxyl side chains and substrate oxygen-containing substituent(s) is quite reasonable, the precedent also raises a question concerning general acid/base catalysts. Structural data for two different malate synthase proteins (13,14) implicate an active site carboxyl side chain in proton abstraction from acetyl-CoA. A recent report of malate synthase mutagenesis experiments supports the importance of the acidic residue (30) but does not clearly establish the true magnitude of the contribution of this side chain. In the case of the HMG-CoA lyase reaction (Scheme 1), substrate cleavage is expected to generate a carbanionic (or enolate) form of acetyl-CoA that must be protonated at the C-2 position to complete the reaction. Thus, the hypothesis that a closely juxtaposed side chain functions as a general acid to support this chemistry seems reasonable. Mutations that eliminate such a side chain should diminish catalytic efficiency by several orders of magnitude. Such effects are observed upon mutation of aspartate 42 (10 3 -fold for D42N; Ͼ10 4 -fold for D42G and D42A) as well as histidine 233 (10 4 -fold for H233A). Both of these residues are conserved in the entire HMG-CoA lyase family of enzymes, which catalyze various reactions that involve proton addition/removal at C-2 of acetyl-CoA. Whereas histidine 233 might function as a general acid at physiological pH, aspartate 42 would require an altered pK to fulfill such a function. Altered pK values for active site carboxyls are well precedented but such a possibility remains to be explored for HMG-CoA lyase. Because there is a significant (Ͼ100-fold) recovery of catalytic efficiency when a side chain with hydrogen bonding potential (D42N) is compared with the observations for a mutant (D42A,D42G) that lacks such potential, it remains possible that D42N can position the side chain of another residue that functions as a surrogate to support (albeit poorly) the normal catalytic function of the carboxyl group of aspartate 42. These various possibilities all argue for an active site location for aspartate 42, which is compatible with the strong correlation of human disease with mutations of this residue.  A) and E72A (panel B) HMG-CoA lyase. Mn 2ϩ binding for E72A was determined in the absence (•) or presence () of the competitive inhibitor 3-hydroxyglutaryl-CoA. The binding of Mn 2ϩ to human HMG-CoA lyase was measured on a Varian Century Line 9 GHz spectrometer with a TE 102 cavity as described previously (10). Each spectrum was recorded at 22°C with a modulation amplitude of 10 G, a modulation frequency of 100 kHz, a microwave power of 60 mW, a field sweep of 1000 G, and a time constant of 0.25 s. The ESR samples contained variable concentrations of Mn 2ϩ (10 -150 M) with constant HMG-CoA lyase sites (50 M). The amount of bound Mn 2ϩ was determined by directly comparing the spectral amplitudes of samples containing HMG-CoA lyase to the corresponding amplitudes observed with a buffered solution containing an equal concentration of Mn 2ϩ in the absence of enzyme. The data were fit by linear regression analysis, and the binding constant (K D ) and stoichiometry (n) were determined from the slope and x intercept, respectively.