Active Site Mapping and Substrate Channeling in the Sterol Methyltransferase Pathway*

Sterol methyltransferase (SMT) from Saccharomyces cerevisiae was purified from Escherichia coli BL21(DE3) and labeled with the mechanism-based irreversible inhibitor [3- 3 H]26,27-dehydrozymosterol (26,27-DHZ). A 5-kDa tryptic digest peptide fragment containing six acidic residues at positions Glu-64, Asp-65, Glu-68, Asp-79, Glu-82, and Glu-98 was determined to contain the substrate analog covalently attached to Glu-68 by Edman sequencing and radioanalysis using C 18 reverse phase high performance liquid chromatog- raphy. Site-directed mutagenesis of the six acidic residues to leucine followed by activity assay of the purified mutants confirmed Glu-68 as the only residue to participate in affinity labeling. Equilibration studies indicated that zymosterol and 26,27-DHZ were bound to native and the E68L mutant with similar affinity, whereas S -adenosylmethionine was bound only to the native SMT, K d of about 2 (cid:1) M . Analysis of the incubation products of the wild-type and six leucine mutants by GC-MS demonstrated that zymosterol was converted to fecosterol, 26,27-DHZ was converted to 26-homo-cholesta-8(9),23(24) E ,26(26 (cid:1) )-trienol as well as

The sterol methyltransferases (SMTs) 1 are appealing targets in the design of enzyme inhibitors, because they lie on a pathway with no counterpart in human physiology (1,2). These enzymes are unique for catalyzing the biosynthesis of 24-alkyl sterol membrane inserts in microorganisms (3,4). Several substrate analogs have been designed to mimic the topological and electrostatic features of the transition state for the C-methylation reaction and several of these compounds, including 24(R,S),25-epiminolanosterol and 25-aminolanosterol (5), markedly inhibit growth of microbes associated with human disease (6 -9).
The SMT from the yeast Saccharomyces cerevisiae has been cloned and overexpressed in Escherichia coli BL21(DE3) cells, and purified to homogeneity (10,11). The enzyme is a tetramer of molecular weight 172,000 and its activity is subject to downregulation by ergosterol and up-regulation by ATP (1). Comparison of protein sequences deduced from the cDNA for fungal SMTs (Fig. 1) revealed a highly conserved region rich in aromatic amino acids, referred to as Region I, containing a signature motif Y81EYGWG86 not present in other AdoMet-dependent methyltransferases. As each of these enzymes uses zymosterol 1A as the sterol acceptor molecule, the aromaticrich domain of Region I has been proposed to be involved in substrate binding, and possibly product formation, by stabilizing intermediate carbenium ions generated during sterol Cmethylation (11). A general stereochemical model (the "stericelectric plug" model) for the coupled Si face (␤-or back face attack) C-methylation of the zymosterol ⌬ 24 -bond 1A and subsequent deprotonation of C-28 with concerted migration of hydrogen to fecosterol 3A has been proposed based primarily on differential inhibition and isotopic labeling studies, as well as recent investigations involving product outcomes resulting from site-directed mutagenesis experiments on SMT action (12)(13)(14).
A central feature of the steric-electric plug model is the predicted intermediacy of a high energy intermediate 2 that can be converted to a ⌬ 24(28) -⌬ 25 (27) -or ⌬ 23(24) -olefinic product (Scheme 1). The usual paradigm for the formation of sterol side chain olefins such as fecosterol from zymosterol in fungi or cyclolaudenol 4B and cyclosadol 5B from cycloartenol 1B in algae and vascular plants, respectively, requires distinct SMTs to generate the variant isomers. However, recent studies with SMTs from fungi and plants employing a range of sterol acceptor molecules and isotopically branching sensitive experiments evoke a common set of enzymatic intermediates involved with the stereochemically related events, electrophilic alkylations, rearrangements, and deprotonations (14,15). Because no gene encoding a protein ⌬ 23 (24) -or ⌬ 25 (27) -SMT activity has been previously identified we were unable to identify consensus motifs that might contribute to the unique regioselectivity of the catalysts. Instead, we performed site-directed mutagenesis of a highly conserved aromatic amino acid residue in Region I, Tyr-81, to establish whether Region I contributes to regiocontrol in the reaction. Unexpectedly, the Y81F mutant exhibited altered substrate specificity catalyzing the successive C-methylation of desmosterol 1C to 24(28)-methylene cholesterol 3C and to a mixture of multiple doubly alkylated sterols found in primitive vascular plants (16), including isofucosterol 7C, fucosterol 8C, and clerosterol 9C. The Y81F mutant SMT failed to generate a mono-alkyl structure other than the ⌬ 24(28) -methylene product from a ⌬ 24(25) -substrate.
A major determinant that controls product diversity is believed to be the precise conformation of the sterol side chain at the sterol binding site (14,17). Although many of the mechanistic and stereochemical details of the C-24 alkylation pathways have been verified (18,19), little is known about the active site of any SMT or the manner in which a SMT enzyme imposes a particular conformation on its acceptor molecule, precisely controls the coupled-methylation deprotonation reaction, and establishes kinetic and substrate specificity for C 1and the successive C 1 -methyltransfers ultimately giving rise to the side chains of fungal ergosterol and plant sitosterol, respectively. We recently reported the first potent mechanism-based inactivator of S. cerevisiae SMT, 26,27-dehydrozymosterol (26,, which made it possible to stoichiometrically and covalently modify the active site of the enzyme (20). A tryptic digest fragment containing the inhibitor-peptide adduct was sequenced and found to be contiguous with Region I. The mechanism of inhibition and covalent attachment of 26,27-DHZ responsible for trapping an active site nucleophile suggested nucleophilic attack of the ⌬ 24 -bond of the S-methyl group of AdoMet leading to a reactive ring-opened intermediate, ⌬ 25(27)olefin 14, shown in Scheme 2, in an analogous manner to the mechanism proposed for the catalysis of 6-cyclopropylidene-3E-methyl-hex-2-en-1-yl-diphosphate for monoterpene cyclase (21).
In this paper, we demonstrate that 26,27-DHZ reacts specifically with Glu-68. The structure of the activated 26,27-DHZ intermediate that leads to enzyme inactivation and the structure of the C-methylation turnover product reveal a novel mode of C-methylation catalysis that is different from the mechanism of irreversible inactivation of monoterpene cyclase by 6-cyclopropylidene-3E-methyl-hex-2-en-1-yl-diphosphate. In addition, we describe product analyses and kinetic results for sitedirected mutagenesis studies that implicate key acidic amino acids in Region I, themselves not required for catalytic activity, capable of directing substrate channeling to produce multiple olefins. % enrichment), and MSD isotopes, and chromatographic materials were as described in our preceding papers (12,21). An antibody to the pure S. cerevisiae SMT was a gift of Dr. M. Venkatramesh at Monsanto Co. and was purified according to the manufacturer's protocol using a serum IgG purification kit supplied by Bio-Rad.

Materials
General Methods-General methods for spectroscopic analysis, heterologous expression, and homogenate preparation and analysis, steady-state kinetic experiments, and protein purification were as described (12,13). The initial velocity data were determined using a Sigma plot 2001 plus Sigma enzyme kinetic program (SSPS Inc). Measurement of K m (app) and V max (app) for sterol employed a concentration range of 5 to 200 M. Data were fitted to the equation, v ϭ V max * (S/K m ϩ S), using a nonlinear least square approach. Kinetic constants possessed Ϯ S.E. of Ϯ5% and R 2 values ϭ 0.95 to 0.97. GC-MS analyses were performed on a Hewlett-Packard 6890 gas chromatograph-mass spectrometer. Routine protein concentrations were estimated according to the method of Bradford with commercial reagents (Bio-Rad) using bovine ␥-globulin as standard (23).
Plasmid Construction of Mutant SMTs-Point mutations were generated using the QuikChange TM site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions, which includes the use of the proofreading thermostable DNA polymerase from Pyrococcus furiosus. Mutagenic oligonucleotides had 23-28-bp overlap with the cDNA sequence before and after the mutated codon, and a melting temperature equal or greater than 78°C (as per Stratagene's instruction manual). Oligonucleotide sequences of mutagenic primers are listed in Table I. The recombinant pET23a-ERG6 was used as the mutagenesis template. Mutations were introduced into bp 161-163, 163-165, 271-273, 306 -308, 315-317, and 363-365 of ERG6 wild type cDNA, to generate six mutant ERG6 proteins. The mutant ERG6 proteins had the following single amino acid changes: E64L, D65L, E68L, D79L, E82L, and D98L. Following DpnI treatment of the mutagenic PCR product, the mutant plasmids were transformed into BL21(DE3) competent cells (from Novagen). All the mutations were verified by automated DNA sequencing to ensure that only the desired mutations and no other changes were inserted into the cDNA.
Expression and Assay of Native and Mutant SMTs-Native and engineered proteins were expressed in BL21(DE3) cells as previously described (14) with minor modifications as follows. Briefly, transformed cells in BL21(DE3) cultures were grown in LB medium containing 50 mg/ml ampicillin overnight. Harvested cells were lysed using a French press, followed by generation of a 100,000 ϫ g supernatant used for analysis of product distribution. The 100,000 ϫ g fraction was purified to apparent homogeneity for K cat determinations as described (14). The initial velocity versus substrate concentration curves for the SMTcatalyzed C 1 -transfer reaction were determined using a fixed [methyl-3 H 3 ]AdoMet concentration of 50 M and varying the concentration of sterol at 5, 10, 15, 25, 75, 100, 150, and 200 M (final concentration). The two substrates dissolved in Tween 80 (1.0%, v/v) were incubated with 0.68 M total protein at 35 o C C 4 for 45 min. The conversion of zymosterol to products was assayed by counting total radioactivity in the nonsaponifiable lipid fraction of the quenched enzyme assay and plotted according to the method of Lineweaver-Burk. Further assessment of the enzyme-generated products was by GC-MS analysis of the nonsaponifiable lipid fraction of activity assays performed overnight at saturating levels of zymosterol and AdoMet.
Detection of SMT Proteins-Aliquots of total and soluble (100,000 ϫ g supernatant) protein from crude E. coli homogenates harboring native or mutant cDNA of the SMT, prepared as described (14), were separated on an 12% SDS-polyacrylamide gel and stained for total protein using Coomassie Blue. The stained gels were scanned with a Amersham Biosciences densitometer, and the density of the isopropyl-1-thio-␤-Dgalactopyranoside inducible band (corresponding to the predicted molecular weight for the SMT protein of about 43,000) was determined as a percentage of the total density of the scan. In addition, Western blot analysis for the expressed proteins using the native SMT antibody was performed under the same chromatographic conditions to estimate SMT levels (14).
The Inhibitor-peptide Adduct-Chemical affinity labeling using pure SMT (0.83 g), AdoMet (100 M), and [3-3 H]26,27-DHZ (100 M) was as described (19). The affinity labeled protein was digested with trypsin and the resulting inhibitor-peptide adduct was separated using a 20% SDS-PAGE gel system. The amino acid sequencing of the isolated radioactive 5-kDa fragment (2.5 ϫ 10 5 disintegrations/min) was carried out on an aliquot (2.5 ϫ 10 4 disintegrations/min) of the excised material using a Porton (Beckman) automated sequencer with on-line Beckman Gold HPLC. Standard Beckman gradient protocols for the micro-phenylthiohydantoin column system were followed for phenylthiohydantoin-derivative separation and identification. Radioactivity of sequencing eluates was determined by liquid scintillation counting (Beckman, LS-6500).
Binding Assay-Determination of the dissociation constant (K d ) for zymosterol and AdoMet was performed by the filtration method, adapted from Ref. 24. Briefly, K d for the two ligands was established using pure native and mutant SMT expressed in BL21(DE3) cells. Assay conditions for AdoMet involved dissolving the ligand in a 1.7-ml reaction vessel containing 100 l of buffer A (50 mM Tris-HCl (pH 7.5), 2 mM MgCl 2 , 1 mM EDTA, 2 mM ␤-mercaptoethanol, and 5% glycerol) at pH 7.5, 0.57 M SMT and incubated at 32°C. For sterol, Tween 80 (1% w/v) was added to the vial to assist solubilizing the sterol. The K d for AdoMet and sterol was determined by aliquoting 0.04 to 3.76 M and 0.25 to 10 M, respectively, to the reaction vessel. After 2.5 h of equilibration on a level shaker, the protein-ligand complex was put on a GF/B glass filter disk (disks were 2.5 cm diameter, Whatman). The filter disks were pretreated with 1 mg/ml bovine serum albumin solution at 4°C overnight to prevent nonspecific binding of the ligand to the disk. Ligand-bound proteins were washed 3 times, 500 M each, with ice-cold buffer A. The disk was air dried and counted by liquid scintillation to determine the amount of radioactivity that remained on the disk from which the amount of bound ligand was determined. Nonspecfic binding was determined in the presence of 100-fold of excess unlabeled AdoMet or sterol. At the specified concentration of ligand, all potential sites of the enzyme are assumed to be saturated and any radioactivity recovered from the assay was viewed to be from nonspecific binding. In no case was nonspecific binding greater than 10%. Binding of AdoMet was independently performed in two 5-cell Scienceware, H420260-0000, equilibrium dialyzers at 32°C. Equilibration conditions employed 1 ml of buffer A, in each chamber, and 0.57 M purified enzyme. The K d value for AdoMet in the presence of SMT was determined by transferring 0.07 to 3 M [methyl-3 H 3 ]AdoMet to the buffer chamber and SMT to the protein chamber. Each side of the cell was separated by Spectrapor membranes with cut-off M r ϭ 6,000 to 8,000. After a 8-h incubation on a level shaker, aliquots were recovered from each chamber, and the radioactivity was measured in a Beckman LS6000 liquid scintillation analyzer to determine the amount of radioactivity in the buffer chamber ([AdoMet] free ) and the protein chamber The equation assumes that there is only one ligand-binding site on the enzyme.

Identification of the Amino Acid Residue Labeled with [3-3 H]26,27-DHZ-
To localize the site affinity labeled with the mechanism-based inactivator in the primary structure, a trypsin digest fragment of about 5 kDa was prepared from SMT that had been treated with [3-3 H]26,27-DHZ. The labeled 5-kDa peptide was subjected to Edman degradation to give an unambiguous sequence of 19 amino acids. The radioactivity associated with the covalently attached irreversible inhibitor was detected between the fifth to tenth cycles (about 80% of the tritium injected into the column was recovered in these fractions) with the major peak of radioactivity eluting in the seventh sequencing cycle (Fig. 2). On the basis of these results, we concluded that Glu-68 was specifically labeled with the inhibitor and represents the putative acidic amino acid residue attached to [3-3 H]26,27-DHZ. However, because the 5-kDa trypsin digest fragment is expected to contain 45 amino acids and six acidic residues at positions Glu-64, Asp-65, Glu-68, Asp-79, Glu-82, and Asp-98 (Fig. 2), it was unclear whether the two additional acidic amino acid residues at positions Glu-82 and Glu-98 not involved in the Edman sequencing of the pep-SCHEME 1 SCHEME 2 tide might also be affinity labeled with the irreversible inhibitor. Therefore, additional experiments were carried out to confirm Glu-68 as the only residue attached to [3-3 H]26,27-DHZ in the SMT.
Effects of Mutations of Acidic Residues on Catalysis and Binding Affinity-In the initial series of experiments it was of interest to establish whether any of the acidic amino acids in the tryptic digest were catalytically dysfunctional. A set of SMT mutants were constructed by site-directed mutagenesis in which the relevant acidic amino residue in the tryptic digest fragment was replaced with leucine. Each of the resulting mutants was expressed in BL21(DE3) cells to similar levels as the wild-type cDNA as revealed by Western blot. The mutant enzymes were subsequently purified in a similar manner to homogeneity, suggesting that none of the mutant proteins were significantly altered conformationally. The steady-state kinetic parameters determined using zymosterol as the sterol acceptor molecule for the mutants are reported in Table II. Although there are differences in the catalytic competence among E86L, D65L, E79L, E82L, and E98L mutants, they were active while E68L was inactive, thereby indicating only Glu-68 is a critical amino acid in the catalysis of 26,27-DHZ. When the amount of enzyme was increased by 10-fold, the E68L mutant continued to fail to exhibit appreciable activity. The rate of catalysis observed for the E98L mutant approached the lower limit of detection in our activity assay system. Nonetheless, we could measure the kinetic constants for 26,27-DHZ using the wildtype SMT, which were K m of 15 M and K cat of 8 ϫ 10 Ϫ4 s Ϫ1 . From our earlier incubations with the recombinant yeast SMT, the k i and K inact of 1.1 M and 1.52 min Ϫ1 , respectively, were determined (20). The partition ratio, a measure of the production of product per inactivation event, can now be calculated from the ratio k cat26,27-DHZ /k inact26,27DHZ , to be 0.03, suggesting 26,27-DHZ to be a potent inhibitor of SMT action.
Binding experiments were performed to definitively confirm that the E68L mutant can bind sterol thereby confirming that the structure of the mutant was not changed significantly by the amino acid replacement. As shown in Fig. 3, zymosterol binds to the native SMT with a K d of about 2 M and 26,27-DHZ binds with a similar K d value (data not shown). Scatchard analysis of the binding data revealed a single binding site for sterol (Fig. 3). AdoMet also binds to the native SMT with similar affinity as the sterols to SMT, K d of 0.5 Ϯ 0.08 M (Fig.  3). Equilibrium dialysis gave a similar binding isotherm for AdoMet generating a K d of 4 M. Based on the two methods, the K d of AdoMet is about 2 M. By comparison, the K d for zymosterol and 26,27-DHZ to the E68L mutant was about 5 M, whereas AdoMet failed to bind at all (no saturation of the enzyme was evident). The leucine substitution appears to generate an impaired active center that prevents productive binding of AdoMet and this may be the reason for inactivity. Because the Y81F mutant can C-methylate fecosterol producing a mixture of 24-ethyl (and ethylidene) sterols (14), fecosterol was tested as a substrate with each of the Leu mutants. In no case was fecosterol a suitable substrate for SMT action.
Identification of Products of Zymosterol Catalysis-The product profiles of the six mutants assayed with zymosterol are summarized in Table II. Each of the active mutants produced fecosterol 3A and in the case of D79L and E82L was accompanied by a second olefin which in GLC eluted earlier than fecosterol (retention time relative to cholesterol of 1.16; molecular weight m/z ϭ 398, 3A) as shown in Fig. 4. The mass spectrum of the new sterol is identical to an authentic specimen of 24-methylzymosta-8,25 (27)  over product-12 in a manner analogous to the GLC behavior of zymosterol to fecosterol (Fig. 4). However, upon further study, we detected two unknown compounds in a 3 to 1 ratio that are generated in 2% yield from the saponification mixture of   monol with a molecular weight of M ϩ 399, suggesting that methyl to methylene transformation failed to occur during the C-methyltransfer reaction. The trideuterated methyl group was assigned to the terminal allylic methyl group at C-26 by the disappearance of the signal resonating at ␦ 1.71 in the 1 H NMR spectrum, which is assigned to C-28 in the product. Thus C-28 is derived from the methyl group attached to AdoMet. The monol is assigned structure 18.  C-28). The COSY and TOCSY correlation indicated connectivity between the proton at ␦ 3.66 ppm and a pair of protons at C-23 (␦ 1.38, 1.52) and C-25 (␦ 2.07 and 2.24), suggesting that the hydroxyl group in the side chain was located at C-24. In addition, the protons associated with C-25 shows connectivity with the exomethylene at C-27 and the allylic methyl group at C-26, suggesting that the "extra methyl group" was attached to C-26 rather to C-24. The mass spectrum of the deuterated diol following incubation with [methyl-2 H 3 ]AdoMet contained diagnostic fragments: M ϩ 417; M ϩ -CH 3, 402; M ϩ Ϫ H 2 O, 399; M ϩ Ϫ 33, 384), consistent with the diol assigned to structure 20 (Scheme 3).
There was insufficient material to establish the stereochemistry of the hydroxyl group at C-24 unambiguously. Therefore, the diol was prepared synthetically from the C-24 aldehyde of zymosterol C-3 acetate as outlined in Scheme 4. Aldehyde 21 was prepared as described in Ref. 27  The Mosher ( 1 H) method was employed to resolve whether the enzyme-generated diol contained the 24␣-or 24␤-hydroxy group (29). The complete proton assignment for both the 2-methoxy-2-phenyl-2-(trifluoromethyl)acetic acid (MTPA) (S)-and (R)-esters of synthetically prepared 20 were determined by extensive two-dimensional NMR analyses (data not shown). The ␦ (␦ S Ϫ ␦ R ) values shown in Scheme 5 for the (S)-and (R)-MTPA indicate that the configuration of the 3-hydroxy and 24-hydroxy groups are ␤-oriented. Because the enzyme-gener-ated diol co-elutes with the synthetic diol used to prepare the MTPA esters, the configuration of the 24-hydroxyl group in the natural product must be ␤-oriented. The D79L and E82L mutants were assayed with saturating amounts of 26,27-DHZ and AdoMet overnight. The resulting product yield and ratio of monol and diol were equivalent to a similar activity assay performed with wild-type enzyme.

DISCUSSION
According to the steric-electric plug model, C-methylation of zymosterol by yeast SMT is to occur via a noncovalent pathway whereby methyl addition to ⌬ 24 and deprotonation of C-28 gives rise to a nucleophilic rearrangement in which H-24 migrates to C-25 on the Re face of the substrate double bond in concert with the initial ionization. Kinetic studies indicate a random Bi Bi mechanism for sterol C-methylation (22). Until recently, few details of protein structure were available that would lend credence to the hypothesized mechanism. It was proposed that a set of aromatic amino acids in the active site would direct folding of the sterol side chain into an appropriate catalytic conformation for C-methylation and stabilize the positively charged high energy intermediate during rearrangement (12,17). In addition, it was proposed that a carboxylate anion would serve as the deprotonating agent of C-28 methyl and concurrently act as an enzymatic counterion to AdoMet (30).
To date, however, no SMT has been crystallized, therefore no three-dimensional structure is available to assist in the analyses. We therefore conducted further affinity labeling using 26,27-DHZ and site-directed mutagenesis experiments with the yeast SMT to determine the precise location of a general base in the active site and to gain information on the role of acidic amino acids in the tryptic digest fragment. Our observation that 26,27-DHZ binds irreversibly to the Glu-68 of yeast SMT suggests the residue is situated in a subsite of the active center. The formation of fecosterol by leucine mutants Glu-64, Asp-65, Asp-79, Glu-82, and Glu-98 but not by the E68L mutant, offers additional support for the involvement of Glu-68 in catalysis of sterol acceptor molecules, including 26,27-DHZ.
The C-methylation of 26,27-DHZ was directional, that is, it was initiated through a backside nucleophilic attack of a methyl group on the methylenecyclopropane leading to ring opening and ring expansion. The termination step gives rise to either olefinic (elimination) or hydroxyl (water addition) products. Ring strain and the rich -electron density associated with the overlapping sp 2 systems of the monosubstituted cyclopropane ring of 26,27-DHZ make the cyclopropanoid susceptible to C-methylation from AdoMet and nucleophilic rupture leading to selectivity with respect to the direction of bond cleavage (Scheme 3). We propose that the C-methylation-elimination reaction involving 26,27-DHZ proceeds by the C-methylation of C-26 of substrate 10 followed by formation of intermediate 16, which can undergo corner to corner nucleophilic rearrangement and cyclopropane ring fission to generate chain extension. The resulting intermediate 17 can be deprotonated to generate 18 (path a), or intermediate 17 can be trapped by an active site nucleophile 19, probably by a nonconcerted ionization of 17 (path b), thereby rendering the protein inactive. Stereospecifc formation of diol 20 with a 24␤-hydroxyl group and the transoid orientation of the 23,24-double bond in 18 during enzyme-catalyzed transformation of 10 provides particularly strong support for the proposition that orientation effects related to the conformation of the side chain in binding or transformation are critical to the effectiveness of substrate analogs as SMT inhibitors (5,14,17). It appears most likely that elongation of the side chain by methylation of C-26 arrested the ongoing intramolecular rearrangements at the terminating carbocation and the final deprotonation of C-23 yielded 26-homo-cholesta-8(9),23(24)E,26(26Ј)-trienol 18. The abortive C-methylation product, diol 20, thus generated by catalysis of 26,27-DHZ and released from the enzyme by saponification is diagnostic of the structure and stereochemistry of the normal enzyme-bound intermediates. A water bridge between the Glu-68 and the cationic intermediate can also form. In which case, formation of diol 20 may result from the addition of water to the cationic intermediate 18. The novel methylenecyclopropane rearrangement catalyzed by SMT reported for the first time is unrelated to cyclopropyl sterol synthesis in marine organisms (31). The mechanism for ring opening of the methylenecyclopropane of 26,27-DHZ is distinctly different from the mechanism involved with methylenecyclopropane inactivators studied in fatty acid metabolism (32).
The detection of the 24␤-OH group in diol 20 suggests that the alkylated enzyme species was formed from the Si face of the original substrate double bond, on the side of the 24,25-double bond where the carboxylate anion should reside for mechanistic reasons. However, because Glu-68 is likely involved with 23deprotonation it must be spatially distal to the nucleophile involved with C-28 deprotonation. The cyclopropane ring of 26,27-DHZ by virtue of its ring strain and electron-donating substituent provides structural features absent from zymosterol that upon catalysis of the acceptor molecule will allow for Glu-68 to be brought into range of this rotationally flexible side chain structure.
The exact nature of Glu-68 involvement in zymosterol catalysis is not clear. The generation of 26-homocholesta-8(9),23(24)E,26(26Ј)-trienol 20 from catalysis of 26,27-DHZ by native SMT suggests that Glu-68 acts as a cryptic base in C-23 deprotonation. The failure of AdoMet to bind to the E68L mutant suggests that the acidic amino acid may interact with AdoMet in the active center of the native enzyme as a counterion. Of interest is that the K d of AdoMet for the SMT was 2 M similar to the K d of 11 M for the AdoMet:glutamyltransferase from Salmonella typhimurium (33).
It seems unlikely that Glu-68 serves as the same active site base responsible for deprotonation in formation of fecosterol from zymosterol because no ⌬ 23 -sterols are formed by the yeast SMT. Given that Glu-68 is a nonconserved residue in fungal SMT sequences and 26,27-dehydrocycloartenol 10B fails as a mechanism-based inactivator of the alga P. wickerhamii SMT indicate the existence of subtle differences between the two groups of SMTs, perhaps because of sequence variations at or near Glu-68. Because the methylenecyclopropane structure is a unique inhibitor for S. cerevisiae SMT, it holds the potential to serve as a lead for the rational design of species-specific agents aimed at modulating SMT activity, with a view to control and/or regulate ergosterol homeostasis in the cell.
The high degree of sequence conservation for the signature motif of Region I in fungi supports a catalytic function for this domain. Region I is also found in plants with a similar consensus sequence (14). Site-directed mutagenesis experiments of the acidic residues of the tryptic digest fragment of the yeast SMT provided additional support that Region I has a functional role in catalysis and that these specific amino acids are not deprotonating agents involved with the conversion of zymosterol to fecosterol. Of the six acidic amino acids in the tryptic digest fragment, only the leucine mutants of Asp-79 and Glu-82 elicited substrate channeling. The Asp-79 and Glu-82 mutants furnished, in addition to fecosterol, a novel C-methylsterol product with the ⌬ 25(27) -olefin structure. The deprotonation leading to the formation of ⌬ 25(27) -olefin involves protons chemically and geometrically distinct from that lost in the generation of ⌬ 24(28) -olefin indicating a critical feature of enzymatic control that can lead to multiple olefin production by a fungal SMT. The isosteric interchange of aspartate or glutamtate with leucine make the active site region less compact. A somewhat loosely packed active center may be more flexible thereby allowing the sterol side chain to develop interactions with polar amino acids that otherwise would not be contacted during catalysis.
In summary, we observed that the yeast SMT active center, normally generating a single olefin, has the requisite amino acids to give rise to multiple olefins that occur naturally in plants. Introduction of the D79L and E82L mutations into the SMT active site results in the formation of a mixture of ⌬ 24(28)and ⌬ 25(27) -olefins as the final products, suggesting that hydrophobic residues in the binding pocket can affect partitioning through a common high energy intermediate 2 in the SMT. The acidic amino acids Asp-79 and Glu-82 must lie on the ␣-face of the sterol side chain at the time of binding otherwise they would abstract the proton from H-24 during the 1,2-hydride shift involved with the coupled methylation-deprotonation reaction catalyzed by SMT. The Glu-68 that interacts with the sterol side chain ␤-face appears to be essential for catalysis but it is not a likely acidic amino acid unit for C-28 deprotonation in zymosterol conversion to fecosterol. Finally, it should be noted that differences in the amino acid composition of Region I that exist in plants compared with fungi may contribute to the different sterol specificities and product mixtures catalyzed by SMT isoforms.