Identification of the Principal Catalytically Important Acidic Residue of 3-Hydroxy-3-methylglutaryl Coenzyme A Reductase*

Kinetic analysis of 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase

Although the activity of mutant enzymes Gl@ + Gln and AsplS3 -, Ala was undetectable under standard assay conditions, their K,,, values for substrates were 4-300-fold higher than those for wildtype enzyme. K, values for wild-type enzyme and for mutant enzymes Glu" + Gln and Aspls3 -* Ala were, respectively: 0.41, 73, and 120 mM ((R,S)-mevalonate); 0.080, 4.4, and 2.0 (4) and rat liver HMG-CoA reductases (5), the P. mevalonii enzyme catalyzes the reversible interconversion of HMG-CoA and mevalonate as well as two half-reactions of mevaldate (3). A 182-kDa homotetramer, P. mevalonii HMG-CoA reductase lacks the "membrane anchor" domain of eukaryotic HMG-CoA reductases. mvaA, the gene that encodes P. mevalonii HMG-CoA reductase, has been cloned, sequenced, overexpressed in Escherichia coli (6), and the protein product has been purified to homogeneity in high yield (3). This laboratory has shown that cysteine residues are nonessential for catalysis, substrate recognition, or substrate binding by P. mevalonii HMG-CoA reductase (7). However, little additional information relative to the catalytic mechanism is available. Apart from early kinetic investigations of the HMG-CoA reductases of yeast (4) and of a pseudomonad (8), few recent publications have addressed the catalytic mechanism of this enzyme. During their investigation of the mechanism of catalysis by yeast HMG-CoA reductase, Veloso et al. (9) proposed, on kinetic grounds, that the formation of mevaldate thiohemiacetal was facilitated by proton transfer from a carboxyl group of the enzyme to the carbonyl oxygen of mevaldate (Fig. 1). The resulting enzyme-carboxylate anion was further proposed to facilitate ionization of the thiol hydrogen of CoASH, thereby facilitating its nucleophilic attack at the carbonyl carbon of mevaldate.
Comparison of the conserved acidic residues within the catalytic domains of 11 HMG-CoA reductases revealed 1 conserved aspartate and 2 conserved glutamate residues (Fig.  2). We therefore used site-directed mutagenesis of the mva gene of P. mevalonii in an attempt to identify the particular acidic residue essential for catalysis by HMG-COA reductase. Corp.), ol-"'S-dATP and a site-directed mutagenesis kit (Amersham Corp.), HMG-CoA agarose and coenzyme A agarose (P-L Biochemicals).

Reductase
Mutants GluS2 + Gln, Glua3 + Gin, AsplR3 -+ Asn, and AsP'~" + Ala-The 3 conserved acidic residues, glutamates 52 and 83 and aspartate 183, were changed to glutamine (Glu"' and Glu""), asparagine (Asp""), or alanine (Asp'=) by site-directed mutagenesis. That each construct contained only the intended mutation was established by DNA sequencing and fragment replacement (see "Experimental Procedures"). Differences in properties between wild-type and mutant enzymes thus reflect mutations at the targeted amino acid residues.

HMG-CoA
Reductase Protein and Enzymic Activity in Crude Cell Extracts-As judged by SDS-PAGE and immunoblotting, all of the mutant HMG-COA reductases were efficiently expressed. Cell extracts were also assayed for their ability to catalyze the oxidative acylation of mevalonate. The specific activity of mutant enzyme Asp"' --, Asn (2.0 pmol of NADH/min/mg) approximated that of wildtype HMG-CoA reductase (2.7 pmol of NADH/min/mg). No activity (specific activity less than 0.02 Fmol of NADH/min/ mg) was detected for mutant enzymes Asp'63 + Ala, G1u5' + Gln, or Glus3 + Gln.

Mutant
Enzymes-Since assays in crude extracts might fail to detect low level activity, we purified all four mutant proteins preparatory to more detailed analysis of their properties.
As judged by SDS-PAGE, each purified enzyme had a subunit size of approximately 43,000 daltons and was more than 90% homogeneous (Fig. 4).

Catalysis by Purified
Wild-type and Mutant HMG-CoA Reductuses-The ability of wild-type enzyme and of all four mutant enzymes to catalyze the standard HMG-CoA reductase reaction (reductive deacylation of HMG-CoA to mevalonate) was measured under conditions optimal for wild-type enzyme (Table II, column 2). Since under these conditions, while mutant enzyme AsplR" + Ala had only 3.6% of wildtype activity, mutant enzyme Aspla3 -+ Asn exhibited an activity 69% that of wild-type enzyme. Aspartate 183 thus is unlikely to be essential for catalysis. Residues Glu5' and GluR3 remained viable candidates since their activities were only 1.8% (Glu"' += Gln) and 0.38% (Glue" + Gln) that of wildtype enzyme. Low activity might, however, reflect a diminished ability to bind one or more substrates. If substrate affinity were unaltered, the activity of wild-type and mutant enzymes would be expected to decrease by equivalent amounts as the substrate concentration was reduced. Activity was therefore next measured under conditions which, for wildtype enzyme, approximate half-saturating conditions for either HMG-CoA or NADH. For mutant enzymes AspIg + Asn and G1uR3 + Gln, the observed decrease in activity was comparable to that for wild-type enzyme (Table II, columns 3 and 4), suggesting that no major change in K, values for these substrates had occurred. Activity was, however, undetectable for mutants Glu"'+ Gln and Aspla3 + Ala, suggesting that these mutants might have altered K, values for substrates. We therefore employed a kinetic assay to determine the K,,, values for substrates in the oxidative ac-  (Fig. 5); for (R,S)-mevalonate (Fig. 6), coenzyme A (Fig. 7), and NAD' (Fig. 8) for mutant HMG-CoA reductases; and for (R,S)-HMG-CoA for mutant enzyme GlP + Gln (Fig. 9). As suggested by the data of Table II, mutants Glu'" --, Gln and AspIs += Asn exhibited essentially wild-type Km values for mevalonate, coenzyme A, and NAD+ (Table III). By contrast, for mutant enzyme AspIs + Ala, K, values were 290-(mevalonate), 25-(CoASH), and 3.8-fold (NAD+); and for mutant enzyme Glu"' ---* Gln, 177-(mevalonate), 55-(CoASH), and 17-fold (NAD+) that for wild-type enzyme (Table III). Clearly, conditions optimal for measurement of wild-type enzyme grossly underestimate the catalytic activity of certain mutant forms of the enzyme. Since V,,, for mutant enzymes Glu"' + Gln, AsplR" + Asn, and Aspls3 + Ala were 15,65, and 109% that for wildtype enzyme, neither AspIs nor G1u5* can be the acidic residue functional in catalysis. However, despite having K,,, values for all three substrates comparable to those for wild-type enzyme, the V,,, for mutant enzyme GlP + Gln was only 0.2% that of wild-type enzyme (Table III). Residue Glu'" thus appears to be the strongest candidate for the acidic residue functional in catalysis.

Mobility of Mutant
Enzymes on Affinity Supports-We suspected from their atypical behavior during chromatography on DEAE-Sephadex that the conformation of mutants Glr.6" -+ Gln and AsplR" + Ala might differ significantly from that of wild-type enzyme. Since the ability to bind substrates provides a measure of structural integrity, we next subjected each mutant enzyme to chromatography on supports that are derivatives of substrates of the reductive deacylation and oxidative acylation reactions of HMG-CoA reductase. Wildtype enzyme, mutant enzyme AsplR3 + Asn, and mutant enzyme Glu'" -P Gln all bound to HMG-CoA agarose (Fig.  10) and to coenzyme A agarose (Fig. 11) affinity supports and could be eluted by 1.0 M KCl. By contrast, enzymes Glu5' + Gln and AspIs + Ala bound to neither support or only poorly to HMG-CoA agarose (AspIs + Ala).

DISCUSSION
A comparison of the derived primary structure of P. mevalonii HMG-CoA reductase with those of the catalytic domains of 11 eukaryotic HMG-CoA reductases revealed 3 conserved acidic residues: Aspls3, G1u5*, and G1ua3. We employed sitedirected mutagenesis, enzyme purification, kinetic analysis, and affinity chromatography to identify GlP as the acidic residue of P. mevalonii proposed by Veloso et al. (9) as functional in catalysis. AS~'~~ and GhP' may be excluded from consideration since amino acid substitutions at these positions yielded enzymes with V,,,,, values from 15 to 109% that of wild-type enzyme. In sharp contrast, the V,,,,, for mutant enzyme Glus3 * Gln did not exceed 0%0.4% that of wildtype enzyme for either HMG-CoA reductase reaction. The drastically lowered catalytic activity of mutant enzyme Glum + Gln does not appear to result from an altered conformation or the inability to bind substrates. Its K,,, values for NAD+, (R,S)-mevalonate, and coenzyme A and its chromatographic behavior on coenzyme A and HMG-CoA affinity supports were typical of wild-type enzyme. We therefore conclude that Glus3 of P. mevalonii HMG-CoA reductase, and by extension the corresponding glutamate residue of the conserved Pro-Met-Ala-Thr-Thr-Glu-Gly-Cys-Leu-Val-Ala motif of all eukaryotic HMG-CoA reductases, is the acidic residue active in catalysis.
Although residues Glu" and AsP"~ are not essential for catalysis, both their strict conservation and the observation that their mutation profoundly affects certain properties suggest that they nevertheless fulfill important functions.
Although precise roles cannot be assigned at this time, G11.3~ and Asp'@ may function in substrate recognition or in maintaining a native conformation.
The failure of mutant enzyme G1u5' + Gln to bind to CoASH and HMG-CoA affinity supports is most simply understood in terms of decreased binding affinity for two substrates that probably share portions of a common binding pocket. The 4.4 mM K,,, for CoASH, a value 55-fold above that for wild-type enzyme (0.08 mM), supports this inference. However, the K, values for NAD' and mevalonate also are elevated by 17-and 177-fold, respectively. Glu'* clearly serves a function imperfectly fulfilled by glutamine. Whether this function involves binding substrates, effects on ordered addition, or perturbation of a domain common to all three substrates cannot be decided at present.
The ability to form a hydrogen bond may be important for As~'~. Mutation of AspIs to asparagine yielded an enzyme with wild-type activity, K,,, values, and behavior on affinity supports. By contrast, its mutation to alanine elevated K, values for all substrates from 4-to 290-fold and impaired the ability to bind to substrate-specific affinity supports. It is nevertheless striking that despite these changes, the V,,, for mutant enzyme AspIs --$ Ala is identical to that of wild-type enzyme. Aspls3 thus may function in substrate binding. The more profound effect upon the K,,, for mevalonate, and secondarily upon coenzyme A, may imply that Aspie resides in the mevalonate-binding pocket or between the mevalonateand CoASH-binding pockets. Alternatively, mutation of Aspla3, like G1u5*, may drastically alter the conformation.