Cellular Sterol Ester Synthesis in Plants Is Performed by an Enzyme (Phospholipid:Sterol Acyltransferase) Different from the Yeast and Mammalian Acyl-CoA:Sterol Acyltransferases *

A gene encoding a sterol ester-synthesizing enzyme was identified in Arabidopsis . The cDNA of the Arabidopsis gene At1g04010 ( AtPSAT ) was overexpressed in Arabidopsis behind the cauliflower mosaic virus 35S promoter. Microsomal membranes from the leaves of overexpresser lines catalyzed the transacylation of acyl groupsfromphosphatidylethanolaminetosterols.Thisactivitycor-related with the expression level of the AtPSAT gene, thus demon-stratingthatthisgeneencodesaphospholipid:sterolacyltransferase (PSAT). Properties of the AtPSAT were examined in microsomal fractions from the tissues of an overexpresser. The

activity (20). We now report that a third of these LCAT-like genes in Arabidopsis encodes a PSAT. We have overexpressed the AtPSAT gene in Arabidopsis behind the 35S promoter, and we characterized the enzyme regarding substrate specificity and selectivity in microsomal membranes from the overexpressers. In addition we investigated the SE synthesizing activity in two T-DNA insertion mutants of the PSAT gene.

EXPERIMENTAL PROCEDURES
Plant Material-Arabidopsis thaliana (ecotype Columbia-0) plants were grown as described earlier (19). Plant materials for gene expression and enzymatic studies were obtained from liquid cultured plants (19). Arabidopsis insertion mutant lines (21) for the AtPSAT gene were identified in the Salk Institute T-DNA insertion library data base (signal-.salk.edu/cgi-bin/tdnaexpress), and seeds were obtained from the Nottingham Arabidopsis Stock Centre (University of Nottingham, UK). Screening for putative mutants was done by PCR (19). Individual plants from each mutant line lacking a T-DNA insertion in the AtPSAT gene (null segregants) were also identified in the same PCR screening.
Isolation of AtPSAT cDNA-Primers designed from the genomic AtPSAT sequence N123456 (corresponds to protein At1g04010) were used to isolate an AtPSAT cDNA fragment using total RNA from 2-week-old Arabidopsis seedling. Poly(A) ϩ RNA was prepared from total RNA with the QuickPrep mRNA micropurification kit (Amersham Biosciences) and used to synthesize single-stranded cDNA with oligo(dT) 18 primer and Moloney murine leukemia virus reverse transcriptase (Amersham Biosciences) by following the standard protocols of Sambrook et al. (22). PCR was performed on this cDNA using the primer 3027s11 (5Ј-TCCTAAACTCTGGCGATG-3Ј), primer 3027a11 (5Ј-CTTTCCCGTAATCCCAGT-3Ј), and AccuTaq (Sigma). The PCR product was diluted 1:40 and reamplified by using AccuTaq and primers 3027s12 (5Ј-AAAGTAGTTCTAGACACCGA-3Ј, the XbaI site is underlined) and 3027a12 (5Ј-CAATCCGCATCTAGAAGTG-3Ј, the XbaI site is underlined). The resulting PCR product was cleaved with XbaI and cloned into Bluescript II SKϪ. PCR was performed on 10 colonies using T3/T7 primers. One clone, pBL-3027, with an insert of the appropriate size was chosen and sequenced.
A missing portion of its 3Ј end was obtained with a modified vectorette PCR method (23) essentially as follows. Total RNA was isolated from Arabidopsis roots grown for 2 weeks in MS liquid media. Poly(A) ϩ RNA was prepared from about 60 g of this total RNA with QuickPrep mRNA micropurification kit (Amersham Biosciences) and used to synthesize double-stranded cDNA with oligo(dT) 18 primers and Moloney murine leukemia virus reverse transcriptase (Amersham Biosciences) by following the standard protocols of Sambrook et al. (22). The cDNA was made blunt-ended and ligated to the preannealed vectorette primers bub1 (5Ј-TCCTCTCCCTTCTCGAATTCTAAGCGGCCGCTC-GAGGATCCCTGTCCTCTCCTTC-3Ј) and bub 2 (5Ј-GAAGGAGA-GGACGCTGTCTGTCGAAGGTAAGGAACGGACGAGAGAAG-GGAGAG-3Ј). Annealing of the vectorette primers was accomplished by mixing the primers (ϳ2 M) in sterile water and heating at 65°C for 5 min, whereupon MgCl 2 was added to give a final concentration of 2 mM, and the mixture was allowed to cool to room temperature. PCR was performed on this cDNA linked to vectorette primers using the primer 3027sen1 (5Ј-GGATCATCACGGATATCATTTATGA-3Ј) and bubspec (5Ј-CGAATTCTAAGCGGCCGCTCGAGGATCCCT-3Ј, the XhoI site is underlined). Amplification was performed with AccuTaq (Sigma), and the conditions used for PCR were as follows: eight identical mixtures were amplified, with a final MgCl 2 concentration of 2 mM, by an initial denaturation at 94°C for 2 min, 35 cycles consisting denatur-ation 94°C, 30 s; a linear gradient where samples varied in annealing temperature from 70 to 60°C; elongation at 72°C for 1 min. After a final elongation cycle at 72°C for 5 min, samples were pooled, and an aliquot was reamplified by using the same conditions with the exception that the primer combination used was the bubspec primer together with a nested primer 3027si (5Ј-GATGAGACGGTACCCTATCATTCA-3Ј, the KpnI site is underlined). The PCR products from this reaction were cloned as XhoI/KpnI fragments into Bluescript II SKϪ, and PCRs were performed on 24 colonies using T3/T7 primers. Appropriate clones were identified by Southern blotting and sequenced. The clone pK3027/ 16-114 was identified having a stop codon and a poly(A) tail. To construct a full-length PSAT cDNA, a KpnI fragment of the At3027 cDNA fragment was cloned into the KpnI site of the plasmid pK3027/16-144 to give the plasmid pK9-126.
Construction of Plant Expression Vector pGVP35S:AtPSAT-To constitutively express the AtPSAT gene, an XhoI/XbaI fragment of pK9-126 was placed downstream to the cauliflower mosaic virus 35S (35S) promoter by cloning it into the XhoI/XbaI site of the binary vector pG35S2 and thereby creating pG35S3027-178 (pGVP35S:At-PSAT). The pG35S2 vector was obtained by replacing the napin promoter in the vector pGPVT-kan (24) with the 35S promoter from pBI121 (Clontech). The construct integrity was confirmed by sequencing.
Transformation of Arabidopsis-Arabidopsis plants were transformed with Agrobacterium tumefaciens GV3101 harboring the binary plasmid pGVP35S:AtPSAT using the floral dip method (25). Homozygous T 3 seed lines expressing AtPSAT as well as their null segregants were identified according to Ståhl et al. (19) and used for lipid and enzymatic analyses.
RNA Isolation and RNA Blot Analysis-Root and leaf tissue of Wt plants grown on agar plates and leaves from T 2 plants transformed with pGVP35S:AtPSAT, were used for RNA extraction. Ten g of total RNA was analyzed for Northern blotting as described by Ståhl et al. (19) using a 2-kb AtPSAT cDNA as probe.
Microsomal Membrane Preparations and Enzyme Assays-Microsomal fractions from leaves and roots of Arabidopsis plants grown in liquid culture were prepared as described previously (27). In order to remove most of the non-polar lipids from the membranes, the microsomal preparations for some enzyme assays were pretreated with acetone as follows. 1 ml of microsomal membranes (corresponding to 0.5-1.5 mol of microsomal PC) in 100 mM potassium phosphate buffer, pH 7.2, was added to 50 ml of cold (Ϫ80°C) acetone under stirring, and the solution was left in a cold room for 10 min with slow stirring. The membranes were collected by centrifugation (5000 ϫ g for 10 min), and the remaining acetone was removed under vacuum for 3 h. The acetone-treated membranes were then resuspended in 1 ml of water and stored in aliquots at Ϫ80°C before being used in enzyme assays.
Aliquots of crude or acetone-treated microsomal fractions (ranging from 5 to 24 nmol of microsomal PC, equivalent to 20 -100 g of protein) were lyophilized overnight. Various radioactive and non-radioactive substrates, added as indicated in the tables and figures, were dissolved in 15-30 l of benzene and then added to the dried microsomes. The benzene was immediately evaporated under a stream of N 2 at 35°C, leaving the lipids in direct contact with the membranes, after which 0.1 ml of 50 mM potassium phosphate, pH 7.2, was added. The suspension was thoroughly mixed and incubated at 30°C. It should be noted that prolonged exposure of the microsomes to benzene severely affected sterol ester synthesis. The incubations were terminated, after periods of times indicated in the tables and figures, by extracting the lipids into chloroform (28).
All enzyme assays were done at least in duplicate, and the amounts of the synthesized radioactive lipids never deviated more than 10% between duplicate samples from the same microsomal preparations. Data shown are standard means of replicates.
Lipid Analysis-Lipids in the chloroform fractions obtained from the assays were separated by TLC with hexane/diethyl ether/acetic acid (70:30:1 by volume) using straight phase Silica Gel 60 plates (Merck). Wax esters and SE co-migrated in the solvent front. To further resolve these lipids from each other as well as to separate SE with different acyl or sterol moieties, the lipids were eluted from the gel by methanol/ chloroform (2:1), extracted into chloroform (28), and rechromatographed on reversed phase high performance TLC (RPHPTLC; Merck) in acetonitrile/tetrahydrofuran (40:60 by volume). The radioactive lipids were visualized and quantified on the plates by using electronic autoradiography (Instant Imager, Packard Instrument Co.). This system separated 16:0/18:1, 18:0, 18:2, and 18:3 cholesterol esters from each other. Also, the dominating wax esters formed by the membranes from 14 C-acyl substrates were clearly separated from the [ 14 C]acyl-cholesterol by the RPHPTLC system. Furthermore, RPHPTLC separated some of the sterol esters with the same acyl group from each other, making it possible to do acyl acceptor competition experiments with the PSAT. Radioactive wax esters were synthesized in assays with all the different 14 C-acyl substrates, but the amount, relative to the radioactive sterol esters formed, was not significant in the membranes from the PSAT overexpresser when cholesterol was used as an acyl acceptor. However, when some sterols that were poor acceptors were used or when assays were done with membranes from Wt or PSAT mutants, co-migration of wax esters and sterol esters was still a problem, even after separation on RPHPTLC. Also, the presence of endogenous sterols in the membranes gave a high "background" of sterol ester synthesizing activity in assays in the absence of added sterols. The formation of radioactive sterol esters and wax esters from 14 C-acyl substrates in the absence of added sterols or fatty alcohols was reduced by 90% after acetone treatment of the microsomal membranes, indicating that most of endogenous sterols and fatty alcohols had been removed by the acetone. Furthermore, the specific activity of PSAT (based on amount of microsomal PC) was increased about 5-fold in the acetone-treated compared with untreated membranes. Quantification of the PC content in the microsomal fractions was done as described earlier (19).
Determination of Sterol Ester Content in Planta-Sterol ester content of the two T-DNA insertion mutants and their null segregants were determined from plants grown in liquid medium. Lipids were extracted from freeze-dried material (100 -200 mg), and sterol esters were purified by TLC and analyzed by gas-liquid chromatography and gas chromatography-mass spectrometry as described previously (29).

RESULTS
Cloning of AtPSAT-In the search for the function of the six PDAT/ LCAT-like genes found in the Arabidopsis genome data bases, we have identified previously the catalytic function of At5g13640 as a PDAT (19) and At3g03310 as a phospholipase A 1 (20). In the present paper we report the identification of the At1g04010 gene as a phospholipid:sterol acyltransferase (AtPSAT). Cloning and sequencing of AtPSAT (Gen-Bank TM accession number AY989885) revealed a gene sequence of 4426 bases containing 15 exons and an open reading frame that encodes a protein of 633 amino acids (Fig. 1A). This protein shows 28% identity with human LCAT. It is, like the AtPDAT but unlike the soluble LCAT, predicted to have one membrane spanning region at the N-terminal end according to TMHMM2.0 (30). It should be noted that the now published amino acid sequence for AtPSAT differs in its 3Ј end from the existing predicted sequence in the GenBank TM .
Sequence Comparison between PSAT and Other LCAT-like Proteins-EST clones from Medicago truncatula (GenBank TM accession numbers BE321377 and BI267156) and Citrus sinensis (GenBank TM accession number CK939714) presenting strong homologies with AtPSAT were identified in data bases. After complete sequencing, these cDNAs were shown to encode proteins of 634 amino acids (MtPSAT) and 641 amino acids (CsPSAT) having 76 and 75% identity, respectively, with AtPSAT. In the frame of the rice (Oriza sativa) genome sequencing, a gene encoding a protein showing 73% identity with AtPSAT was also found.
Alignments were performed for deduced amino acid sequences of the following cDNAs (Fig. 1B): a first group of plant PDATs, a second group containing AtPSAT and the three orthologues named MtPSAT, CsP-SAT, and OsPSAT, genuine mammalian and avian lecithin cholesterol acyltransferases (LCAT), and various LCAT homologues. Isolation and characterization of plant cDNAs have been described elsewhere (20). Six characteristic conserved regions are shown in Fig. 1B. From these alignments, it is clear that the catalytic triad (Ser-Asp-His) already found in mammalian LCATs (31), plant and yeast PDATs (19), and plant phospholipase A 1 (20) is also present in Arabidopsis PSAT and putative PSATs. This triad is closely associated with the catalytic mechanism of LCAT enzymes in which a fatty acid is cleaved from the sn-2 position of phosphatidylcholine and trans-esterified to the free hydroxyl group of cholesterol to generate cholesterol ester. Site-directed mutagenesis at Ser-195, Asp-461, and His-546 of AtPSAT should be performed to validate this hypothesis in the case of plant PSATs.
An important conserved hydrophobic domain (domain 1 from left to right in Fig. 1B) is found in all LCAT-like proteins. Yet its function is unknown. A second domain of interest ( 152 AVPYDYRLSP in AtPS-AT) is highly conserved in all sequences of Fig. 1B. This domain might be involved in phospholipid binding for the following reasons. (i) All proteins listed in Fig. 1B use a phospholipid as substrate. (ii) It has been shown recently (32) that Glu-173, a residue of human LCAT situated inside this domain, is the molecular determinant for substrate specificity of HsLCAT for PC containing 18versus 20-carbon sn-2 fatty acyl chains. (iii) It has been recently suggested that the amphipathic helix encompassing residues 175-198 in the human LCAT, which is very close to the domain of the PSAT considered here, could be involved in PC binding in human LCAT (8). It is clear that a directed mutagenesis study of this domain (Ala-152 . . . in AtPSAT) should give important information concerning substrate specificity (nature of the phospholipid substrate and of the fatty acid transesterified). Remarkably, this domain begins with an alanine only in enzymes that transfer a fatty acid from a phospholipid to a sterol (mammalian and avian LCATs and plant PSAT). Finally, the fourth domain (Fig. 1B) shows a conserved sequence Gly-X-Pro-X (where X is Leu, Val, or Ile) restricted to animal LCATs, plant PSATs, and the group of plant proteins called LCAT1, the function of which is still unknown. Such a domain might be a candidate for sterol binding.
Functional Characterization of AtPSAT-To characterize the catalytic function of AtPSAT, we overexpressed the At1g04010 gene in Arabidopsis by using the constitutive 35S promoter. Three transgenic lines showing different levels of expression on Northern blotting as well as a null segregant (Wt) of the line with the highest expression were selected for further experiments. Expression of the AtPSAT was below the detec-tion limit on Northern blotting in Wt, whereas electronic autoradiography of the blots of the transformant lines D28-6-2-, D28-3hyphen]4-8 and D28-1-5-gave relative expression levels of 1:2:8, respectively ( Fig. 2A).
Plants from the different lines were grown in liquid culture, and microsomal preparations from the leaves were tested for SE synthesizing activity. Assays were performed with sn-1-16:0-sn-2-[ 14 C]18:2phosphatidylethanolamine (PE) in the absence and presence of cholesterol. The amount of radioactivity found in the SE area upon separation by TLC was small in the Wt and not dependent on the addition of cholesterol. The transgenic lines showed radioactive SE formation correlating with the expression levels of the AtPSAT gene and with 3-fold stimulation by addition of cholesterol (Fig. 2B). The activity in D28-1-5-, the highest expresser, was more than 40-fold higher than in the Wt. Assays were also performed with [ 14 C]cholesterol at different concentrations (Fig. 2C). The SE formation was just above detection limit in the Wt, whereas the three transgenic lines produced radioactive SE amounts correlating with the expression level of the AtPSAT gene and with an increasing amount produced with increasing concentrations of added cholesterol. Thus, it is apparent that AtPSAT encodes an SE-synthesizing enzyme. The accumulation of SE continued in a non-linear fashion over a period of at least 2 h in the membranes from the highest overexpresser (data not shown).
Radioactive monoacylglycerols, diacylglycerols, and free fatty acids were formed in detectable amounts in incubations of Arabidopsis microsomes with [ 14 C]18:2-PE as substrate (data not shown). Since it has been reported that neutral lipids could serve as acyl donor for SE synthesis (13)(14)(15)(16)(17), we investigated if these lipids were more immediate acyl donors than PE for the SE-synthesizing enzyme. SE and wax esters could be separated from each other on RPHPTLC (see "Experimental Procedures"). From these separations it was evident that that no cholesterol esters were formed from the radioactive neutral lipids, but a radioactive compound with higher polarity was synthesized in assays with membranes from both control and overexpresser. Addition of 22:1 ⌬13 fatty alcohol resulted in the formation of 4 -6-fold more radioactivity in a polar compound co-migrating with authentic linoleoyl-docosaenol esters and with wild type and overexpresser accumulating similar data (data not shown).
On the basis of these findings, we conclude that PE, but not neutral lipids, could serve as an immediate acyl donor and cholesterol as an acyl acceptor for the AtPSAT protein. We here suggest the name phospholipid:sterol acyltransferase or PSAT for this plant enzyme.
Lipid and Positional Specificity of the PSAT-The AtPSAT activity in microsomal fractions from the AtPSAT overexpresser was examined with three different phospholipids, PC, PE, and PA. The activity toward PE was 5.5-fold higher than PC and over 30 times higher than PA (TABLE ONE). We also compared the activity toward the different sn  (TABLE ONE). When corresponding incubations were done with microsomal preparations from Wt, radioactive SE was below the detection limit in any of the assays with [ 14 C]PC as the substrate (TABLE ONE).
Sterol Specificity of the AtPSAT-The specificity of the AtPSAT toward different sterols and sterol intermediates was tested in microsomal preparations from the AtPSAT overexpresser. The experiments were first performed with crude microsomal fractions, and a number of the sterols were shown to have an inhibitory effect in the production of SE compared with assays without the addition of sterols (data not shown). In order to remove as much as possible of the endogenous sterols in the membranes, we pretreated the microsomal preparations with acetone. When AtPSAT was assayed in the acetone-treated microsomes without addition of cholesterol, the activity was decreased to 10 -20% that of assays with the untreated microsomes. Moreover, the amount of wax esters produced in the acetone-treated microsomes was reduced by over 90% compared with the untreated microsomes (data not shown). Assays in order to determine acyl acceptor specificity were performed with acetone-treated microsomal preparations from both roots and leaves of the AtPSAT overexpresser (Fig. 3B). Various sterols and precursors were tested, as well as two close triterpenes (lupeol and ␤-amyrin), which do not belong to the sterol biosynthesis pathway but to an adjacent pathway. Cycloartenol, 24-methylene cycloartanol,  obtusifoliol, and 24-ethylidene lophenol are intermediates in plant sterol biosynthesis, leading to the main end products campesterol, sitosterol, and stigmasterol, whereas cholesterol is a minor (Ͻ5%) end product sterol in Arabidopsis and in most plants. Lanosterol, zymosterol, and ergosterol are yeast sterols. Lupeol and ␤-amyrin appeared not to serve as substrates for the AtPSAT. Among the 14 tested sterols and sterol intermediates added, it was only the cholesta-5,7-dienol that did not significantly increase the activity over background. The activity toward the two major sterols in Arabidopsis, campesterol and sitosterol (33), was about half that of dihydrocholesterol, zymosterol, and cholesterol, the three best substrates. Although the relative activity for different sterols showed the same trend in both root and leaf membranes, there was a clear difference for certain sterols. The experiment was repeated once with different batches of microsomal preparations with essentially the same result. This indicates that the observed specificity was not entirely due to an inherent property of the enzyme per se but that it was also influenced by the nature of the membrane to which it was attached. It should be noted that the SE synthesis was below the detection limit for all sterols in assays with acetone-treated membranes from Wt plants (data not shown).
Sterol Selectivity of the AtPSAT-Some of the sterols, particularly the intermediates in the biosynthesis of the end product sterols, were quite poor acceptors for the AtPSAT (see above), and yet they are known to accumulate as SE in some plant tissues such as Apium graveolens cell suspension culture, Avena sativa roots, Gossypium hirsutum buds and anthers, and a sterol over producer tobacco mutant (4). We therefore performed a sterol selectivity study of the AtPSAT. Some of the acylated sterol intermediates could be separated from the acylated end product sterols by RPHPTLC. For example, the intermediates 18:2-cycloartenol, 18:2-obtusifoliol, and 18:2-24-ethylidenelophenol clearly separated from 18:2-sitosterol. When the intermediates were presented as single substrates, the PSAT activity for these substrates was less than half that for sitosterol. However, when presented as equimolar mixtures with sitosterol, the acylation of sitosterol was decreased by up to 80%, and the activity toward the intermediates increased up to 3-fold (Fig. 4). The total amount of sterol acylated was relatively unchanged compared with the assays with only sitosterol. A series of incubations with sitosterol and cycloartenol, where one compound was held at constant concentration and the amount of the other was increased, are shown in Fig. 5, A and B. Although addition of increasing amounts of cycloartenol to sitosterol decreased overall sterol ester formation (Fig. 5A), the addition of increasing amounts of sitosterol to cycloartenol increased overall sterol ester formation as well as acylation of cycloartenol (Fig. 5B).
AtPSAT Activity in T-DNA Insertion Mutants-Two putative AtPSAT insertion mutant lines were obtained from the SALK collec-tion. Analysis of these lines revealed a T-DNA insertion in the sixth exon (SALK_117091) and in the first intron (SALK_037289) (Fig. 1A). Both mutant plants showed a similar and strong reduction in their total SE content (about 70%) when compared with that of Wt (TABLE  TWO), indicating that AtPSAT is the major SE-synthesizing enzyme in Arabidopsis. However, the residual SE content indicates leaky mutants or additional enzymes existing able to acylate sterols. Experiments were therefore performed in order to biochemically characterize the remaining SE synthesizing activity in Arabidopsis microsomal preparations. Because SE synthesizing activity was just on the limit of detection in Wt in our previous experiments, we now used substrates with about 10-fold higher specific radioactivity. Because the RPHPTLC separated cholesterol esters with 18:0, 16:0/18:1, 18:2, and 18:3 from each other, we could study the acyl selectivity of the SE-synthesizing enzymes from endogenous acyl donors in the membranes from both mutants and Wt, using radioactive cholesterol as acyl acceptor. The SE synthesizing activity in microsomal preparations from the mutants, the Wt, and the overexpresser showed the same pattern of acylation of cholesterol with mainly 18:3 and 18:2 incorporated and minor amounts of 18:0 and with the mutants accumulating about half the amounts of SE compared with the Wt (Fig. 6A). Because the microsomal preparations synthesized a range of wax esters, some of which co-migrated with the SE on RPHPTLC plates, it was not possible to determine the SE synthesis from radioactive PE as a donor in crude microsomal fractions from Wt and mutants. However, acetone-treated microsomes accumulated much less wax   (Fig. 6B). In assays with acetone-treated microsomes and radioactive cholesterol, the mutants showed extremely low, but yet detectable, formation of SE (Fig. 6C).
From these experiments it can be concluded that the AtPSAT activity that we can measure in acetone-treated Wt microsomes is below the detection limit in the mutants and that the remaining SE in these plants are most probably synthesized by another enzyme. This enzyme appears to use a different acyl donor but to have similar acyl specificity as the AtPSAT. This enzyme is either severely inactivated in acetonetreated microsomes or the acyl donor substrate for enzyme is removed during acetone treatment. However, acyl-CoA is not a substrate for this reaction.

DISCUSSION
In this work we have identified the first plant gene encoding an SEsynthesizing enzyme and further characterized the encoded enzyme from Arabidopsis. We show that the enzyme catalyzes the transacylation of acyl groups from phospholipids to a variety of different sterols, and we suggest the name phospholipid:sterol acyltransferase (PSAT) for this type of enzyme. For several reasons, we prefer not to call this enzyme LCAT, despite its catalytic similarities with this animal enzyme. First, lecithin is an old name for PC, and the plant enzyme prefers PE over PC as an acyl donor. Furthermore, cholesterol is by far not the only acyl acceptor for the plant enzyme. It should also be noted that the animal enzyme is an extracellular enzyme and plays another physiological role than PSAT.
Relatively little information on SE biosynthesis in plants is available in the literature. Published work indicates that the synthesis involves transacylation and that both phospholipids and neutral lipids, in particular diacylglycerols and triacylglycerols, could serve as acyl donors (13)(14)(15)(16)(17). We could not detect any activity toward neutral lipids by the PSAT enzyme. However, we found that our microsomal preparations rapidly hydrolyzed the neutral lipids to free fatty acids that were acylated to endogenous fatty alcohols to produce wax esters. These wax esters comigrate with SE on silica gel chromatography. It is therefore possible that some of the previous work in sterol ester biosynthesis, using enzyme assays with radioactive neutral lipids as acyl donors, did not discriminate wax ester synthesis from sterol ester synthesis.
The PSAT had a strong preference for PE over PC, hardly utilized PA as substrate, and did not use neutral lipids as acyl donor. This suggests that the enzyme has quite strict requirements for the type of polar head group of the phospholipids. However, it does not necessarily mean that the binding site for the phospholipid substrate is highly specific. It could also be that the enzyme has easier access to the bilayer "disturbing" PE molecules than the bilayer forming phospholipids. However, it is relevant here to compare the PSAT with a structurally related enzyme, the PDAT, which is forming triacylglycerols from phospholipids and diacylglycerols (18,19). The Arabidopsis PDAT has only 1.7-fold specificity for PE over PC (compared with 5-fold for the PSAT) and utilizes also PA at significant rates (19). Moreover, it was inferred that the AtPDAT, to a lesser extent, could also utilize diacylglycerol as an acyl donor (19). Therefore, it appears that PSAT has inherent higher acyl donor specificity than the PDAT.
All common acyl groups present in plant membranes were utilized well by the PSAT enzyme, although specificity for polyunsaturated fatty acids was seen. The enzyme was highly specific for position sn-2, which in plant lipids is essentially devoid of saturated acyl groups. This suggests that PSAT enzyme would play a minor role in the synthesis of SE with saturated fatty acids.
The sterol specificity of PSAT is broad, and both intermediates and end product sterols were acylated. However, no activity could be seen with the triterpenes lupeol and ␤-amyrin, which suggests that another enzyme is responsible for the formation of esters of these compounds. This is also consistent with the observation that the ester amount of these triterpenes was not reduced in the T-DNA insertion mutants (34). Among the various plant end product sterols, the rate of acylation was in the order cholesterol Ͼ campesterol Ն sitosterol Ͼ stigmasterol. This is in good agreement with the composition of the SE pool of a number of species where the proportions of cholesterol and stigmasterol are higher and lower, respectively, than that of the free sterol pool, whereas the proportions of campesterol and sitosterol are similar in both pools (4).
Some of the intermediates in sterol biosynthesis were poor substrates when presented as single substrates but were utilized preferentially and at a high rate when sitosterol, the major sterol in Arabidopsis, was present. The remarkable stimulation of acylation of sterol intermediates in the presence of an end product sterol suggests that the PSAT could be  an allosteric enzyme, positively regulated by end product sterols but not by the intermediates. It should be noted that the catalytic activity for the intermediates and end product sterols was similar, provided that the enzyme was activated by end product sterols. It can be hypothesized that such an unusual regulation of enzyme activity could work in vivo as a modulator of the flux in the sterol biosynthetic pathway as follows. When end product sterols reach a threshold level in the membranes, the PSAT enzyme will be fully activated at its allosteric binding site and will efficiently esterify intermediates in SE biosynthesis and thereby slow down the further accumulation of end products. When the level of end product sterols drops to a point where PSAT is no more activated, the flux through the pathway will be resumed. Such an hypothesis finds some support from results obtained in a sterol overproducing tobacco mutant. The pool of free sterols was quantitatively and qualitatively relatively unchanged in the mutant compared with Wt, but the pool of SE was increased 10-fold and contained both intermediates and end product sterols (35). A similar sterol profile was observed in transgenic Arabidopsis overexpressing the early enzyme in the sterol biosynthesis pathway, the hydroxymethylglutaryl-CoA reductase (29). Indeed, hydroxymethylglutaryl-CoA reductase is likely to be responsible for the regulation of the total flux in the biosynthesis of sterols (36), whereas PSAT contributes to maintain homeostasis of free sterols in the membrane as well as to limit the amount of free intermediates in sterol biosynthesis, which otherwise could be deleterious for membrane functions (4).
The two T-DNA insertion mutants studied had no detectable activity with PE as acyl donor when measured with acetone-treated microsomes, suggesting a lack of active PSAT enzyme in these plants. Yet these mutants were still able to synthesize SE, although the levels were strongly reduced (with about 70%) in intact plants. This indicates that PSAT is the main SE-synthesizing enzyme in Arabidopsis. The ubiquitous expression of the PSAT gene (see www.plantbiology.msu.edu/lipids/genesurvey/) reinforces this assumption. We could show that the acyl preference, using radioactive cholesterol and endogenous acyl donors present in the membranes, was similar in microsomal preparations from Wt and mutant and also similar to that of overexpressed PSAT. Because SE synthesizing activities in microsomal membranes from Wt and mutants were extremely low and formation of wax esters interfered with the detection of SE using radioactive acyl donors, a more detailed biochemical characterization of the remaining SE synthesizing activity in the mutants was not possible. It should be noted that the specific activity of the PSAT enzyme in the wild type Arabidopsis (measured as pmol ϫ min Ϫ1 ϫ mg protein Ϫ1 in microsomal fractions) was less than one-fifth that of the Arabidopsis PDAT enzyme (19).
The PSAT enzyme is a member of the PDAT/LCAT family. Indeed, the alignment of the amino acid sequences of the SE-synthesizing enzymes, the animal LCATs, AtPSAT, and other putative plant PSATs, with the yeast and plant triacylglycerol-synthesizing PDAT enzymes, reveals that important sequence similarities exist among these proteins. As discussed above, some of these domains might be involved in substrate binding and/or catalytic mechanism. A directed mutagenesis study should clarify this point.
In summary, the first plant gene encoding an SE-synthesizing enzyme, PSAT, has been identified and shown to belong to the LCAT/ PDAT gene family. This is the first identified intracellular enzyme catalyzing the synthesis of SE by an acyl-CoA-independent reaction. Thus, the physiological function of this enzyme is totally different from the evolutionarily related animal LCAT. PSAT is likely to serve similar cellular functions as the unrelated ACAT enzymes, found in animals and fungi. The PSAT enzyme utilizes a wide range of sterols and sterol intermediates. Esterification of sterol intermediates is greatly stimulated by presence of end product sterols, suggesting a role for PSAT in regulating both the amount and the quality of the free sterols in the membrane. The in vivo effects of overexpressed PSAT and mutations in the PSAT gene are currently under investigation. Taken together, although our results suggest the presence of an additional enzyme(s) able to acylate sterols, they clearly show that PSAT is the major SE-synthesizing enzyme in Arabidopsis.