Function-Structure Studies and Identification of Three Enzyme Domains Involved in the Catalytic Activity in Rat Hepatic Squalene Synthase*

Rat hepatic squalene synthase (RSS, EC 2.5.1.21) contains three conserved sections, A, B, and C, that were proposed to be involved in catalysis (McKenzie, T. L., Jiang, G., Straubhaar, J. R., Conrad, D., and Shechter, I. (1992) J. Biol. Chem. 267, 21368–21374). Here we use the high expression vector pTrxRSS and site-directed mutagenesis to determine the specific residues in these sections that are essential for the two reactions catalyzed by RSS. Section C mutants F288Y, F288L, F286Y, F286W, F286L, Q293N, and Q283E accumulate presqualene diphosphate (PSPP) fromtrans-farnesyl diphosphate (FPP) with reduced production of squalene. F288L, which retains approximately 50% first step activity, displays only residual activity (0.2%) in the production of squalene from either FPP or PSPP. Substitution of either Phe288 or Phe286 with charged residues completely abolishes the enzyme activity. Thus, F288W, F288D, F288R, F286D, and F286R cannot produce squalene from either FPP or PSPP. All single residue mutants in Section A, except Tyr171, retain most of the RSS activity, with no detectable accumulation of PSPP in an assay mixture complete with NADPH. Y171F, Y171S, and Y171W are all inactive. Section B, which binds the diphosphate moieties of the allylic diphosphate subtrates, contains four negatively charged residues: Glu222, Glu226, Asp219, and Asp223. The two Glu residues can be replaced with neutral or with positively charged residues without signficantly affecting enzyme activity. However, replacement of either Asp residues with Asn eliminates all but a residual level of activity, and substitution with Glu abolishes all activity. These results indicate that 1) Section C, in particular Phe288, may be involved in the second step of catalysis, 2) Tyr171 of Section A is essential for catalysis, most likely for the first reaction, 3) the two Asp residues in Section B are essential for the activity and most likely bind the substrate via magnesium salt bridges. Based on these results, a mechanism for the first reaction is proposed.

The biosynthesis of hepatic cholesterol is an essential and complex process that is not yet fully understood. In mammals, the first specific step in the pathway is the conversion of trans-farnesyl diphosphate (FPP) 1 to squalene, a reaction catalyzed by squalene synthase (farnesyl-diphosphate:farnesyl-diphosphate farnesyltransferase (EC 2.5.1.21) (Fig. 1).
An active, truncated form of the rat hepatic enzyme was the first mammalian squalene synthase (RSS) to be purified (1). It is a membrane-bound enzyme localized to the endoplasmic reticulum (2). The cDNA for RSS, cloned from a cDNA library, was successfully expressed in Escherichia coli (3). The fulllength native protein, determined from the cDNA sequence, contains 416 residues. Using similar methodology, the human hepatic squalene synthase was cloned, and its transcriptional regulation in HepG2 cells was studied (4). This enzyme was also cloned in yeast by complementation (5).
The sequence identity between mammalian species is 90%, whereas the identity between rat and yeast is only 44.8%. Comparison of the rat and yeast proteins revealed that three regions, referred to as Sections A, B, and C, were highly conserved and therefore likely to be essential for catalysis (3). Sequences of high homology to Section A are also present in phytoene synthase from various species (13,14).
Section B was first identified as an aspartate-rich motif in a variety of enzymes that utilize prenyl diphosphates as substrates, including phytoene synthase. This motif was proposed to bind the diphosphate moiety of the allylic substrates via a magnesium salt bridge (15)(16)(17)(18)(19).
Unlike Sections A and B, Section C of RSS has not been identified in any other proteins, including those using FPP as a substrate. This section or one highly similar to it is present in all known squalene synthase enzymes from different species (Fig. 2). The 19-residue sequence is fully hydrophobic and does not contain any charged residues. It was proposed to consist of a hydrophobic helix at its C terminus and a hydrophobic ␤-sheet at the N terminus. These two structures are separated at Pro 292 in the mammalian enzymes (3).
Much of the interest in squalene synthase has been stimulated by the unique chemistry of carbon-carbon bond formation. The formation of squalene from FPP takes place via two separate reactions that represent unique and important steps in the formation of lipophilic cholesterol intermediates (Fig. 1). In the first reaction, the condensation of two molecules of FPP forms a cyclopropyl-containing intermediate, presqualene diphosphate (PSPP), through a unique 1Ј-2-3 prenyl transferase re-action. PSPP is then rearranged and reduced by NADPH in the second step to form squalene (15, 20 -23).
Phytoene is a squalene-like intermediate in the biosynthesis of carotenoids, which are produced by plants and certain bacteria. Phytoene is also synthesized in two steps (5,24) (Fig. 1). In the first reaction, two molecules of geranylgeranyl diphosphate (AAPP) are condensed head-to-head to form a cyclopropyl ring-containing intermediate, prephytoene diphosphate (PPPP). This reaction is similar to the first one of squalene biosynthesis. However, in the second reaction, prephytoene diphosphate is converted to phytoene by a nonreductive rearrangement in which the diphosphate is eliminated. Thus, unlike squalene synthase, this enzyme does not require NADPH for its activity and, therefore, is not expected to contain specific NADPH binding sequences. It was also reported that, as with squalene synthase, the two steps leading to the formation of phytoene from geranylgeranyl diphosphate are catalyzed by a single enzyme (25).
In the present study we have used site-directed mutagenesis to assess the importance of various residues in the RSS-conserved sequences. We have also investigated the possible involvement of Section C, which is unique to RSS, in catalyzing the second step of the reaction.

EXPERIMENTAL PROCEDURES
Materials-The expression plasmid pRSS1327, an E. coli expression vector for RSS, was originally constructed by McKenzie et. al. (3) and was used in these studies. This construct was expressed in the DH5␣ strain of E. coli (Stratagene). A VCS-M13 interference resistant helper phage was also from Stratagene. For high expression of RSS, the pTrxFus vector was used together with its host cell E. coli strain GI698 (Invitrogene). Sculptor in vitro mutagenesis system and ⌬Taq Cycle Sequencing kits were purchased from Amersham Pharmacia Biotech; [1-3 H]FPP (143 Ci/mmol) and nonlabeled trans-FPP were from NEN Life Science Products; restriction enzymes and T4 ligase were from New England Biolab; 125 I-protein A (30 mCi/mg) was from Amersham; squalene was from ICN Biochemicals.
Construction of High Expression Vector for RSS-Two RSS expression vectors were used, pRSS1327 and pTrxRSS. pRSS1327, a pBluescript based vector, expresses a 52-kDa chimeric protein consisting of ␤-galactosidase fused to the full-length RSS protein (4 kDa of ␤-galactosidase and 48 kDa of RSS) and was constructed as described previously (3). pTrxRSS was formed by subcloning of the RSS cDNA sequence in pRSS1327 into pTrxFus, utilizing the KpnI at the 5Ј end and the BamHI at the 3Ј end as cloning sites. Thus, pTrxRSS expressed an active fusion protein with a molecular mass of 60 kDa (12 kDa portion of thioredoxin at the N terminus and 48 kDa of RSS sequence at the C terminus). Under the same conditions, pTrxRSS expresses a 6 -9-fold higher enzymic activity than pRSS1327 (see Fig. 3) and, therefore, was used as the preferred expression vector. However, since single strand DNA can not be rescued from pTrxRSS, we have utilized pRSS1327 as a shuttle vector for site-directed mutagenesis.
Expression and Preparation of RSS-pRSS1327 was expressed in the E. coli strain DH5␣ as described previously (3). Tryptophan (10 g/ml) was used for the induction of the expression of pTrxRSS in the GI698 strain of E. coli. After 4 h induction, cells were harvested by centrifugation at 5,000 ϫ g for 5 min at 4°C and washed once with TGED buffer (50 mM Tris-HCl, 5% glycerol, 10 mM MgCl 2 , 5 mM EDTA, 1 mM dithiothreitol, pH 7.9). Cells were lysed with lysis buffer (50 mM Tris-HCl, 1 mM EDTA, 10 mM dithiothreitol, 5 mM CHAPS, 1.5 mg/ml lysozyme, pH 7.9) at room temperature for 30 min. Cell lysates were then stored at Ϫ80°C overnight. After 18 h, cell lysates were quickly thawed and sonicated at 4°C. The 5,000 ϫ g supernatant of this RSS preparation was immediately used for further analyses. The protein concentration of RSS preparations was determined by the Bradford method using BSA for calibration (26).
SDS-Polyacrylamide Gel Electrophoresis-SDS-polyacrylamide gel electrophoresis was performed by the method of Laemmli (49) in a mini-gel system (Bio-Rad) at 100 volts. 10% acrylamide gel was used for the separation of proteins. Pre-stained protein standards (Bio-Rad) were used for molecular weight calibration.
Immunoblotting-Western blots of RSS were performed as described by Shechter et al. (1) with some modifications. After separation by SDS-polyacrylamide gel electrophoresis, the proteins were transferred FIG. 1. Comparison of the consecutive steps in the reactions catalyzed by squalene synthase and phytoene synthase. The first reaction step for the two enzymes is similar (left scheme, squalene synthase; right scheme, phytoene synthase), producing an intermediate containing a cyclopropyl ring. The second reaction step differs in that squalene synthesis, but not phytoene synthesis, proceeds through an NADPH-utilizing reductive rearrangement reaction.

FIG. 2.
Putative sequences in the catalytic site of RSS. The three most highly conserved regions of amino acid sequences between RSS and other known squalene synthase proteins are depicted. Sections A and B also show high similarities to the CrtB gene product, phytoene synthase (PS), which, like squalene synthase, catalyzes head-to-head (1Ј-1) condensation. Identical sequences are indicated by the shaded areas. The positions of the first and last amino acids in each sequence are shown as well as the biological source. ERWHE, Erwinia herbicola; ERWUR, Erwinia urodevora.
to a nitrocellulose membrane (0.22 m) at 4°C for 1.5 h at 100 V. The membranes were then blocked with 5% dry milk in PBST buffer (2.03 g of NaH 2 PO 4 , 11.49 g of Na 2 HPO 4 , 85 g of NaCl, and 20 ml of Tween 20 in a total volume of 1000 ml) overnight at room temperature. The blocked membranes were rinsed with PBST buffer and incubated with rabbit anti-RSS antiserum diluted 1:1000 in PBST buffer with 1% bovine serum albumin for 1 h. The membranes were washed with PBST buffer three times for 5 min each rinse. Radiolabeling was performed by incubating the membranes for 1 h with 125 I-protein A (30 mCi/mg) diluted 1:1000 in PBST buffer containing 1% bovine serum albumin. The membranes were then washed three times in PBST buffer and air-dried. Immunoblots were visualized by autoradiography.
RSS Enzyme Actvity Assays-Enzyme assays were performed as described previously (1). Briefly, RSS was assayed in 96-well micro well plates. Each reaction contained 100 mM potassium phosphate, pH 7.4, 5.0 mM MgCl 2 , 5.0 mM CHAPS, 10 mM dithiothreitol, 2.0 mM NADPH, 20 M (0.04 Ci) [1-3 H]FPP, and 20 g of RSS protein in a total volume of 100 l. Enzyme and substrate were incubated for 20 min at 37°C, after which the reactions were stopped by the addition of 10 l of 1.0 M EDTA, pH 9.2. After adding 10 l of 0.5% squalene as a carrier, half of each reaction mixture (60 l) was analyzed by TLC chromatography. Silica gel TLC plates were developed with 5% toluene in hexane, allowing the solvent front to migrate 11 cm. Squalene was visualized by exposure to iodine vapors, and the stained bands was scraped off and counted for radioactivity. Enzyme activities are expressed in pmol of squalene formed/min/mg of protein.
Site-directed Mutation-The construct used for the generation of the mutations is the pBluescript phagemid pRSS1327, which contains the cDNA of the entire coding region of RSS (3). First, single strand DNA was prepared from pRSS1327 using the Sculptor in vitro Mutagenesis kit (Amersham) with minor modifications. Briefly, TG1 host cells were transformed with pRSS1327 to obtain fresh transformants. On the following day, one colony was selected and grown in 2ϫ yeast tryptone medium for 3 h. VCS-M13 helper phage (Stratagene) was then added to the culture, and the incubation was continued for 4 h. The cells were then lysed, and the resulting single strand DNA was purified by PEG 8000 precipitation followed by phenol extraction and ethanol precipitation.
Oligonucleotide-directed mutagenesis was performed as described previously (27). To overcome the high GC content in the mutation site, annealing and extension of mutation primers were performed at a relatively high temperature. Therefore, the following procedure was employed. Single strand DNA and mutation primer were denatured at 94°C for 5 min and slowly cooled to 25°C to allow annealing. The annealed primer was then extended in a buffer containing 2 mM dithiothreitol, 6 mM MgCl 2 , 0.2 mM each of the dNTP mixtures containing dCTP␣S instead of dCTP at room temperature for 10 min. 400 units of T4 DNA ligase were added, and the reaction was incubated at 37°C for 1 h to allow for extension of the mutated strand and ligation. The remaining single strand oligonucleotide was removed by T5 exonuclease. The non-mutated strand was cut by NciI and digested by exonuclease III. The resulting mutated pRSS1327 DNA was used to transform XL-1 Blue competent cells. Single colonies were selected, and mutations were confirmed by sequencing. The mutated RSS cDNA was then subcloned into pTrxFus to form the expression construct pTrxRSS. The presence of the correct mutations in the pTrxRSS vector was again confirmed by sequencing. The synthetic antisense oligonucleotides that were used in this procedure and the various single amino acid mutations that were generated are as follows: Q293N, 5Ј-CGTAGCAATGG- Preparation, Separation, and Identification of PSPP-PSPP was biosynthesized and separated by the method described by Zhang and Poulter (28), except that the enzyme used here was generated by the expression of the vector pTrxRSS. Briefly, pTrxRSS, which contains the wild-type RSS cDNA, was expressed in the host E. coli strain GI698, and the enzymic extract was prepared as described above. A typical squalene synthase reaction was carried out, except that NADPH was omitted from the reaction. EDTA was added to a final concentration of 1 M to terminate the reaction, and the mixture was sedimented at 5000 ϫ g for 5 min. The clear supernatant was subjected to separation by radio-HPLC on a C-18 reversed phase column (Hewlett Packard Liquid Chromatogram 1090 equipped with Packard Radiomatic Floone). For the separation, the column was pre-equilibrated with 25 mM (NH 4 ) 2 CO 3 , and elution of products was accomplished by a 35-min linear concentration gradient to 100% acetonitrile, which was followed by a 5-min linear concentration gradient to 50% tetrahydrofuran. Both radioactivity and UV signals were monitored. A radioactive peak with a retention time reported for PSPP (28) was collected and was further analyzed and identified in a bioassay and by mass spectroscopy.
Large scale preparation of In the bioassay, the collected sample of [ 3 H]PSPP was used as a substrate for the reaction. The assay was done under the same conditions described for the RSS enzymic assay, except that [ 3 H]FPP was replaced by 5 nCi of the isolated [ 3 H]PSPP. The radioactive product was confirmed as squalene by a chromatographic comparison with an authentic sample of squalene using the radio-HPLC separation system described above.
For electrospray mass spectrometric analysis, the sample was first

FIG. 3. Comparison between RSS activities obtained by expression of pTrxRSS and pRSS1327 in E. coli.
Expressions of the two vectors in newly transformed clones were compared under the same conditions. Western blots (A) and specific RSS activity (B) are shown. Immunoactivity was detected by Western blot as described under "Experimental Procedures." pTrxFus represent lack of immunoreactive protein in cells transfected with a vector control. Lane 2 was loaded with 10 g of protein from E. coli and DH5␣-tranfected with pRSS1327, and lane 3 was loaded with 10 g of protein from strain GI698 transfected with pTrxRSS. The molecular size for both were determined by comparison to prestained molecular weight markers. Enzyme activities were determined by radiochemical assay as described under "Experimental Procedures." Three separate assays were performed for each determination of specific activity. In this and subsequent figures, the results are presented as means Ϯ S.D. Value labels in the figure are percentages of the specific activity of RSS from pTrxRSS. lyophilized and then was reconstituted with acetonitrile, 25 mM ammonium bicarbonate (7:3) and infused at 2 l/min flow to the mass spectrometer for analysis. All mass spectra were recorded on Sciex API III triple quadruple instrument equipped with an electrospray source at negative mode. For the molecular mass measurement, the mass spectrometer was scanned in 0.33-Da steps of 3 ms dwell time at unit resolution over the m/z range of 450 -750. For structural elucidation, the collision-induced dissociation experiment was performed on the molecular ion of interest, m/z 585. Argon was used as the collision gas at a thickness of 1.8 ϫ 10 14 molecules /cm 2 . The fragmentation was achieved at a collision energy of 30 eV. The mass analyzing quadruple Q3 for the fragment ion spectrum was scanned from m/z 30 to 600 in 0.33-Da steps of 2 ms dwell at unit resolution. The final spectra were the result of 50 -100 scans.
Determinations of Partial Reaction Activities of Squalene Synthase-The activity of the first step reaction of RSS ( Fig. 1) was assayed as described above, except NADPH was omitted. The product of the first reaction, [ 3 H]PSPP, was quantitatively analyzed by radio-HPLC. Specific activities are expressed as the mean of triplicate reactions.
The activity of the second step was assayed using PSPP and NADPH as substrates. Each reaction contained 5 nmol of [ 3 H]PSPP prepared as described above. The product of the reaction, squalene, was quantitatively analyzed by radio-HPLC. Specific activities are expressed as the mean of triplicate reactions.
Since several mutants were found to accumulate PSPP even in the presence of NADPH, additional reactions were performed under the conditions described above for the determination of RSS enzymic activity, except that instead of using TLC for the analysis, radioactivity of both PSPP and squalene were quantitatively assayed by radio-HPLC. The activity of the first reaction in these mutants is calculated as total conversion from FPP to both PSPP and squalene. In all calculations, the molar specific radioactivity of PSPP and squalene was assumed to be 1.5 times that of the [1-3 H]FPP from which it was produced on the basis of the known loss of one proton from the C-1 position of one of the two reacting FPP molecules (29).

Expression of RSS in pTrxFus
Vector-Previously we have reported the expression of RSS in E. coli using pRSS1327, a pBluescript-based vector (3). However, in the present study, where analysis of mutant RSS enzymes often required detection of low level activities, it was necessary to obtain higher levels of RSS expression than is possible with pRSS1327. For this purpose, we have used the pTrxFus vector to construct the E. coli pTrxRSS expression system. The pTrxRSS system yielded up to 9-fold higher enzymic expressions than pRSS1327, as shown in Fig. 3. Therefore, it was used for the mutant RSS enzyme activity analyses. However, a single strand DNA needed for the mutation procedure cannot be rescued from pTrxRSS, and therefore, the original pRSS1327 vector was utilized as a shuttle vector for the site-directed mutagenesis.
Analysis and Purification of Recombinant RSS Reaction Products-The separation, purification, and analysis of the recombinant RSS reaction products was similar to the previously described procedure used for the yeast enzyme (28). The radioactive product was identified as squalene based on its co-migration with an authentic standard of squalene on thin layer chromatograpy and HPLC systems as described previously (1, 3). Its chromatographic migration in comparison to an authentic standard of squalene in HPLC is shown in Figs. 4, A and B. Three lines of evidence show that the intermediate produced in the RSS enzymatic reaction lacking NADPH is PSPP. 1) It has elution properties in the HPLC system identical to the previously described yeast enzyme-derived PSPP (data for the preparation of the yeast-derived PSPP is not shown) (28). 2) Electrospray mass spectroscopy shows a typical m/z 585 for C 30 H 51 O 7 P 2 (M Ϫ H ϩ ) Ϫ and m/z 607 for C 30 H 50 O 7 P 2 Na (M ϩ Na ϩ Ϫ H ϩ ) Ϫ as was reported for this structure (28).
3) The isolated RSS-derived [ 3 H]PSPP (Fig. 4C) could be converted quantitatively to squalene by the same enzyme in a reaction containing NADPH (Fig. 4D). These three criteria together with the observation that PSPP is formed as the major product from FPP in a reaction lacking NADPH (Fig. 4C), unequivocally prove its structure to be that of PSPP.
Mutational Analysis of Section C-Of the three conserved protein sections of squalene synthase, Section C is unique to this enzyme. Therefore, this section was proposed earlier to be directly involved in the catalytic process (3). We report here the single residue modifications of the four amino acids in this section: Phe 286 , Phe 288 , Gln 283 , and Gln 293 .
The two Phe residues, which are at the N-terminal part of this section, are located within a sequence that was predicted to have a hydrophobic ␤-sheet structure (3). We have examined the effect of modifying these two residues on RSS activity by replacing them with aromatic (Tyr or Trp), aliphatic hydrophobic (Leu), or charged (Asp and Arg) residues.
All the resulting mutants retained the typical RSS antigenicity and could be analyzed by Western blots using specific anti-RSS antibodies. As expected, all mutant enzymes retained the 60-kDa size (Fig. 5A). The specific activity of the expressed wild-type RSS was determined to be 1570 pmol/min/mg. Substituting charged residues for either Phe 286 or Phe 288 produced mutant proteins with only residual RSS activity or no activity at all. The mutants F288D, F288R, and F286D did not retain any activity, whereas the calculated specific activity for F286R is 5.4 pmol/min/mg, which is approximately 0.3% that of the wild-type RSS.
Mutations in which alternate aromatic residues or hydrophobic aliphatic residues were introduced had a differential effect on the two Phe positions. Although these substitutions caused substantial loss of activity when introduced to position 286, between 10.4 and 28.6% activity could still be observed. Thus, the mutants F286Y, F286W, and F286L retained specific activities of 449, 266, and 163 pmol/min/mg, respectively. If instead, the same substitutions were introduced into position 288, a much more pronounced effect on the loss of RSS activity was observed. Activities of the mutants F288Y, F288W, and F288L were only 1.6, 1.0, and 0.2% that of the wild-type RSS (Fig. 5B).
In all structurally known squalene synthase proteins, there is a Gln moiety in the middle of Section C located at the most N-terminal position of a predicted ␣-helix structure. In the mammalian enzymes it is Gln 293 . Also, in most but not all of the known enzymes there is a Gln residue at the N terminus of this section terminating the predicted ␤-sheet sequence. In the mammalian enzymes it is Gln 283 (see Fig. 2 and Ref. 3). These two amino acids were mutated to either a shorter residue (Gln 3 Asn) or to a negatively charged residue (Gln 3 Glu). Again, as for the Phe mutants, these also retain the RSS immunogenicity. As expected, the size of the mutated RSS protein is not modified (Fig. 6A). Shortening of these two residues by a single methylene carbon resulted in loss of activity, with Q283N retaining 52.6% (590 pmol/min/mg) and Q293N retaining 9.9% (111 pmol/min/mg) of the wild-type RSS activity. Mutation of either of these residues to a Glu moiety caused an almost complete loss of enzymic activity with less than 3% remaining (15.6 and 29.0 pmol/min/mg for Q293E and Q283E, respectively).
Partial Reaction Analysis Shows That Mutations in Section C Predominantly Affect Reaction 2 of RSS-Wild-type recombinant RSS, like any other native squalene synthase, does not accumulate the intermediate PSPP when NADPH is present. However, since several mutations in this section affect the enzymic activity of RSS only partially, we have studied the two reactions separately with these mutants.
Analysis of the second reaction step shows that the mutation F288L caused almost a complete loss of this activity. Other RSS

FIG. 4. Isolation and identification of [ 3 H]PSPP produced by wild-type RSS from pTrxRSS-transfected cells. [ 3 H]Squalene and [ 3 H]PSPP were prepared biosynthetically from [ 3 H
]FPP using RSS generated by the expression of the vector pTrxRSS in the host E. coli strain GI698. For the production of [ 3 H]PSPP, a typical squalene synthase reaction was performed, except that NADPH was omitted. The reaction was terminated by the addition of EDTA to a final concentration of 1 M, and the reaction mixture was sedimented at 5000 ϫ g for 5 min. The clear supernatant was subjected to separation by radio-HPLC on a C-18 reversed phase column (Hewlett Packard liquid chromatogram 1090 equipped with Packard Radiomatic Flo-One). For the separation, the column was preequilibrated with 25 mM (NH 4 ) 2 CO 3 eluted with a 35-min linear gradient of acetonitrile (0 -100% v/v) followed by a 5-min linear gradient of tetrahydrofuran (0 -50% v/v). Both radioactivity and UV signals were monitored. The radioactive peak of PSPP was collected and identified in a bioassay and by mass spectroscopy. mutants such as F288Y, F286Y, F286W, F286L, Q293N, Q293E, Q283N, and Q283E still retain partial second step activity (Fig. 7B). Interestingly, all mutant enzymes that retain the first activity, including F288L, F286Y, F286W, F286L, Q293N, Q283N, and Q283E, accumulate significant amounts of PSPP even in the presence of NADPH. Among them, a reaction with the mutant F288L results in an accumulation of the highest amount of PSPP, whereas almost no squalene production is detected (Fig. 7C).
The products of the two RSS reactions for the various mutant enzymes were determined and quantitated by radio-HPLC based on their migration properties. For F288L, in which the remaining activity is unexpectedly almost exclusively that of reaction one, a more detailed analysis was performed. The major product of the reaction from [ 3 H]FPP, which eluted in the HPLC at the position expected for PSPP, was collected and analyzed. Similar amounts of [ 3 H]PSPP were produced in this reaction regardless of the presence of NADPH (compare Figs. 8,  A with B). Both of these products were confirmed to be PSPP by electrospray mass spectroscopy, showing a typical m/z of 585 as was reported for this structure (28,30).
The [ 3 H]PSPP produced by F288L in the presence of NADPH (Fig. 8A) was reincubated with the wild-type RSS, also in the presence of NADPH. The intermediate was completely converted to [ 3 H]squalene (Fig. 8C), indicating that this mutation affects the activity of the second reaction almost exclusively.
Mutational Analysis of Section A-Section A of RSS is very similar not only to the same section in squalene synthase from different species but also to a protein sequence in phytoene synthase (Fig. 2) that catalyzes a similar first step of the reaction (Figs. 1). Therefore, we have examined residues in Section A that may be essential for catalysis.
At first, we mutated individual residues bearing either polar or charged groups, since these are most likely to be involved in the catalysis. Tyr 171 and Tyr 174 were mutated to Phe residues individually, Cys 172 was replaced by Ala and Ser, His 173 was modified to Gln, and Arg 185 was mutated to Gly. Expression of these mutant RSS proteins in E. coli using several different expression vectors showed that all except Y171F had retained partial or full RSS activity (data not shown). Therefore, we have studied Tyr 171 , which exists in Section A of all known squalene syntase and phytoene synthase proteins, and compared it to Tyr 174 , which is present in squalene synthase but is replaced by His in phytoene synthase (Fig. 2). These two Tyr residues were further mutated to either Ser or Trp. These mutants along with Y171F, Y174F, and R185G were expressed using the pTrxRSS construct, and their RSS activity and immunoreactivity was compared. All mutations in Tyr 171 resulted in a complete loss of RSS activity, whereas mutations in Try 174 or Arg 185 still retained either partial or even full activity (Fig. 9B). Analysis of activities of the individual reaction steps of RSS indicate that mutants in Tyr 174 and Arg 185 retain partial or even full activities in both the first and second reactions steps of RSS (Fig. 10, A and B). Mutants in Tyr 171 including Y171F, Y171S, and Y171W do not retain detectable first-step activity but do show a residual activity of approximately 2.9% for the second step. This result was reproducibly observed for Y171W (Fig. 10B). In contrast to mutants in Section C of RSS, none of the reactions catalyzed by the Section A mutants accumulate PSPP during the conversion of FPP to squalene in the presence of NADPH (Fig. 10C).
Mutational Analysis of Section B-A randomly isolated clone of pTrxRSS was found to expressed the RSS protein with a specific activity of approximately 50% that of the original wildtype enzyme. This change in activity was not accompanied by any change in protein size or immunoreactivity as determined by Western blot analysis. Sequencing of the cDNA of this clone showed a spontaneous mutation in nucleotide 664 from a G to A. The spontaneous mutation of the codon GAA to an AAA resulted in a modification from Glu 222 to Lys 222 .
The negative charges in Section B were proposed earlier to be involved in the binding of the disphosphate moieties of prenyl diphosphate substrates via magnesium salt bridges. This binding motif is present in a variety of enzymes that act on prenyl diphosphate substrates (15)(16)(17)(18)(19). Therefore we considered it essential to study the effects of mutating Glu 222 , Glu 226 , Asp 219 , and Asp 223 shown to be present in Section B (3) on the activity of RSS. The mutant E226K, which, similar to the spontaneous E222K described above, there is a replacement of a negative charge with a positive one, retained about 90% activity of the wild-type enzyme. In contrast, any attempt to mutate the two Asp residues in this section, including mutations with shorter residues (D219E, D223E), with neutral residues (D219N, D223N), or with a positively charged residue (D223K), resulted in total loss of activity (Fig. 11). Several attempts to prepare D219K mutant enzyme failed. DISCUSSION In this work, the importance of specific residues in three highly conserved sequences of squalene synthase has been assessed by site-directed mutagenesis. The three sequences were first identified by comparing the proteins from yeast and rat (3). Subsequent cloning of the gene from a wide variety of sources including plants has shown that these sections are indeed conserved in all species.
Of the three sections, section B was first identified in FPP synthase as an aspartate-rich domain and was proposed to serve as the diphosphate moieties binding sequences (15)(16)(17)(18)(19). The direct involvement of Sections A and C in catalysis could not be determined. Comparison between known proteins showed that sequences similar to those of Sections A and B are found in all squalene and phytoene synthetases, whereas the sequences in Section C are unique to squalene synthase. (Fig.  2). The possible catalytic function of residues in these two sections can be determined through comparision of the enzyme mechanisms of the two enzymes. The formation of the cyclopropyl intermediate in the first reaction is the same in each. However, the second reaction is different for the two enzymes. In squalene synthase there is a reductive rearrangement of the cyclopropyl intermediate to the linear product utilizing NADPH, whereas in phytoene synthase no reduction is involved in the rearrangement (Fig. 1). Thus, it is reasonable to assume that Section C, which is unique to squalene synthase, may be involved in the NADPH-requiring step.
The evidence presented here supports such possibility. In contrast to the reaction by native enzyme or the recombinant wild-type enzyme expressed by pTrxRSS, enzymes that are mutated in Section C accumulate significant amounts of PSPP under normal reaction conditions (Fig. 7). This observation indicates that mutation of residues in Section C modifies the reaction so that the second step becomes rate-limiting. The most pronounced effect is observed for F288L. Whereas more than 50% activity of the first reaction is retained, only a residual production of squalene is detected from either FPP of PSPP. This significant first step reaction activity was confirmed by the production and the identification of PSPP in this mutant (Fig. 8). Both Phe 288 and Phe 286 are conserved in Section C of all squalene synthase proteins. A radical mutation in which there is an introduction of charge to these positions completely eliminates the activities of both the first and the second reactions, as was observed for F286D, F286R, F288D, and F288R (Fig. 7). However, in more conserved substitutions in which hydrophobic or aromatic residues are introduced, significant second step activities are retained (F286Y, F286W, F286L, and F288Y). The lack of this activity in F288W and F288L may indicate a different function for the two Phe residues in the catalysis.
The most pronounced difference is between F286W, which retains much of the first step and some of the second step activities, and F288W, in which none of the RSS activities are detected (Fig. 7B). Most of the second step activity is retained for Q293N and for Q283N, indicating that the length of the residue in this position may not be crucial for the catalysis (Fig.  7B). Since Q293E and Q283E show significant reduction but not elimination of second step activity, a negative charge there may result in a less favorable enzyme conformation for the catalysis (Fig. 7B).
As indicated above, Section A of RSS is very similar to the comparable section in squalene synthases from different species but also to a protein sequence in phytoene synthase (Fig.  2). Since all polar or charged residues in Section A except Tyr 171 could be mutated with retention of varying levels of RSS activities, the indication may be that these residues are not directly involved in the catalysis. Mutation of Tyr 171 to the mutants Y171F, Y171S, and Y171W completely abolished formation of PSPP or squalene from FPP (Figs. 10, A and B). The same mutation in Tyr 174 had little effect on the total RSS activity (Fig. 10B). This is perhaps expected since Tyr 171 of the mammalian squalene synthases is conserved in the same section of squalene synthases from yeast and plant sources as well as in phytoene synthase enzymes. However, Tyr 174 in that position is unique only to squalene synthase and is substituted to His in the same section in phytoene synthase (Fig. 2).
Analysis of the partial activities shows that the above mutants of Tyr 171 are unable to carry out the first reaction. Limited second reaction activity is observed for Y171F and Y171W (Fig. 10). This result indicates that the phenyl ring is specifically required for activity and cannot be substituted with just either an aromatic or hydroxyl group.
The negatively charged residues in Section B of RSS (Glu 222 , Glu 226 , Asp 219 , and Asp 223 ) are not functionally equivalent. The two Glu residues can be mutated to a positively charged Lys residue with relatively high retention of activity. Thus, it can be assumed that they are not essential for catalysis or for the binding of a diphosphate moiety of one of the reactants. In contrast, it is equally apparent that the two Asp residues are essential. Not only does neutralization or reversal of charge (D219N, D223N, and D223K) cause inactivation, so does the subtle modification of adding a single carbon atom to the side chain (D219E and D223E) (Fig. 11).
The results presented here are consistent with the participation of Asp 219 and Asp 223 of Section B, Tyr 171 of Section A, and Phe 288 of Section C in the binding of the substrates or in the catalysis. Furthermore, based on our observation, it is likely the Tyr 171 is involved in the first reaction, whereas residues in Section C contribute primarily to the catalysis of the second reaction either by providing favorable enzyme conformation or by the binding to, and/or enabling, the proper alignment of NADPH.
The present results also have implications concerning the detailed chemical mechanism of catalysis by squalene synthases. The uniqueness of the reductive coupling of two FPPs to form squalene via PSPP has elicited much mechanistic speculation over the years (15,20,(31)(32)(33)(34)(35)(36). It seems highly probable that carbocation intermediates are involved in both the formation of PSPP and its reductive rearrangement to squalene. The postulation of carbocation intermediates in the first stage is based on analogy to other reactions of prenyl diphosphates, such as that catalyzed by prenyl transferase (15, 22, 37). For FIG. 9. Tyr 171 , located in Section A, is essential for the activity of RSS. Wild-type pTrxRSS and mutant RSS were expressed in E. coli strain GI698 and compared under the same conditions in newly transformed clones. Western blots (A) and specific RSS activity (B) are shown. Immunoactivity was detected by Western blot as described under "Experimental Procedures." Immunoblots of extracts of cells transfected with pTrxFus, wild-type pTrxRSS, and the indicated mutants show a signal of RSS, when present, at 60 kDa. Each lane was loaded with 5 g of proteins. Molecular size was determined by comparison to prestained molecular markers. Enzymic activities were determined in a radiochemical bioassay as described under "Experimental Procedures." Data in the figure are means Ϯ S.D. of specific activities from three independent determinations. The figure shows that of all the mutants in Section A, those of Tyr 171 lack activity.
FIG. 10. Partial reaction activities of mutants in Section A of RSS. Wild-type and mutants were expressed in E. coli strain GI698 and compared under the same conditions in extracts obtained from newly transformed clones. The first reaction activity (A) was obtained in a standard assay from [ 3 H]FPP, as described under "Experimental Procedures," except that NADPH was omitted. The product, [  . Data in the figure are means Ϯ S.D. of specific activities from three independent assays. The results indicate that the three mutants Y171F, Y171S, and Y171W completely lack the first or the total activities (from FPP), and only residual second-step activity (from PSPP) is observed. the second stage, there is good evidence for the intermediacy of carbocations (28). Providing stabilization for such high energy species is, presumably, an important part of the catalytic mechanism of squalene synthases. Therefore, it seems reasonable to assume that sections A and C have evolved to interact favorably with the carbocation intermediates in the reactions 2 ϫ FPP 3 PSPP and PSPP 3 squalene, respectively.
Over the last few years it has been clearly established that one very important mode of stabilization of cations in biological structures is through their interaction with the electrons of aromatic rings (38,39). This cation-interaction has been well documented for ammonium and metal ions, and similar stabilization of carbocation intermediates has been proposed for several terpenoid polyene cyclases on the basis of the location of aromatic amino acid residues in their structures (40 -46). In a very recent theoretical study of the cyclization of squalene oxide to lanosterol, energies of roughly 8 -10 kcal/mol were computed for the stabilization afforded carbocation intermediates by complexation with aromatic residues (47).
The genetic mutations described herein show that aromatic amino acid residues are essential for both stages of the squalene synthase reaction. A possible interpretation of these observations is shown in Fig. 12. The site-directed mutagenesis results strongly suggest that Tyr 171 is playing a key role in the cyclopropane ring formation that occurs during 2FPP 3 PSPP. This, perhaps, is the result of its possessing both an aromatic ring for complexation and an acid-base functional group. Although, as noted earlier, there has been considerable speculation, very little is known about the details of the mechanism of formation of PSPP. The mechanistic pathway shown involves an initial ionization of one FPP to form an allylic carbocation followed by nucleophilic attack by the 2,3 double bond of the other FPP to form an intermediate tertiary carbocation, as shown in the conversions A 3 B 3 C. Abstraction of the pro-S proton (48) from the allylic methylene group then leads to PSPP (C 3 D). If Tyr 171 is appropriately situated, as in A of Fig. 12, it could facilitate initial ionization by both the complexation of the incipient carbocation and by serving as a proton source. This in turn would lead to the intermediate ion pair shown in B. Then a nucleophilic attack on the allylic cation by the double bond of the second, and also proximate, FPP can form a tertiary carbocation, as shown in C. The phenolate anion of Tyr 171 could then serve as a relatively strong base to break the C-H bond and form the cyclopropane ring in PSPP (D).
In the second stage, the conversion of PSPP to squalene is proposed to be catalyzed by residues in Section C. The replacement of either Phe 288 or Phe 286 reduces enzyme activity, suggesting that either or both of these residues could be involved in stabilizing a carbocation intermediate(s). Of the two, the effect of mutation of Phe 288 is more profound. Stabilization of the tertiary carbocation in structure C could be afforded by complexation (not shown in Fig. 12) with Phe 288 (or perhaps also with Phe 286 ). Since a cation is formed at this same carbon atom during the conversion of PSPP to squalene, this same type of interaction could also be obtained in the second stage of the reaction. This would be consistent with the effects of mutating Phe 288 (or Phe 286 ) on the enzyme activity. These mechanistic considerations are obviously speculative, but they are consistent with the current results, and they can serve to explain the singular, critical role of Tyr 171 . Experiments to test the validity of the ideas presented herein are in progress. FIG. 11. Asp 219 and Asp 223 , located in Section B, are essential for the activity of RSS. Wild-type pTrxRSS and the indicated mutant RSS were expressed in E. coli strain GI698 and compared under the same conditions in newly transformed clones. Enzymic activities were determined in a radiochemical bioassay, as described under "Experimental Procedures." Data in the figure are mean Ϯ S.D. of specific activities from three independent determinations. Only residual activity is observed for D219N and D223N, in which a neutral Asn is substituted for the Asp resudues, whereas total loss of activity is observed for D219E and D223E, in which negative charge is retained in the shorter residue, Glu. D223K, which introduces a positive charge in that position, also lacks RSS activity. The initial ionization of one FPP to form an allylic carbocation is followed by nucleophilic attack by the 2,3 double bond of the other FPP to form an intermediate tertiary carbocation, as shown in the conversions A 3 B 3 C. Abstraction of the pro-S proton from the allylic methylene group then leads to PSPP (C 3 D). If Tyr 171 , located in Section A of the enzyme, is appropriately situated as shown in A, it could facilitate initial ionization by both the complexation of the incipient carbocation and by serving as a proton source. This would lead to the intermediate ion pair shown in B. Then a nucleophilic attack on the allylic cation by the double bond of the second, and also proximate, FPP can form a tertiary carbocation, as shown in C. The phenolate anion of Tyr 171 could then serve as a relatively strong base to break the C-H bond and form the cyclopropane ring in PSPP (D).