Cloning and Expression of a Wheat (Triticum aestivum L.) Phosphatidylserine Synthase cDNA OVEREXPRESSION IN PLANTS ALTERS THE COMPOSITION OF PHOSPHOLIPIDS*

We describe the cloning of a wheat cDNA (TaPSS1) that encodes a phosphatidylserine synthase (PSS) and provides the first strong evidence for the existence of this enzyme in a higher eukaryotic cell. The cDNA was isolated on its ability to confer increased resistance to aluminum toxicity when expressed in yeast. The sequence of the predicted protein encoded by TaPSS1 shows homology to PSS from both yeast and bacteria but is distinct from the animal PSS enzymes that catalyze base-exchange reactions. In wheat, Southern blot analysis identified the presence of a small family of genes that cross-hybridized to TaPSS1, and Northern blots showed that aluminum induced TaPSS1 expression in root apices. Expression of TaPSS1 complemented the yeast cho1 mutant that lacks PSS activity and altered the phospholipid composition of wild type yeast, with the most marked effect being increased abundance of phosphatidylserine (PS). Arabidopsis thaliana leaves overexpressing TaPSS1 showed a marked enhancement in PSS activity, which was associated with increased biosynthesis of PS at the expense of both phosphatidylinositol and phosphatidylglycerol. Unlike mammalian cells where PS accumulation is tightly regulated even when the capacity for PS biosynthesis is increased, plant cells accumulated large amounts of PS when TaPSS1 was overexpressed. High levels of TaPSS1 expression in Arabidopsis and tobacco (Nicotiana tabacum) led to the appearance of necrotic lesions on leaves, which may have resulted from the excessive accumulation of PS. The cloning of TaPSS1 now provides evidence that the yeast pathway for PS synthesis exists in some plant tissues and provides a tool for understanding the pathways of phospholipid biosynthesis and their regulation in plants.

We describe the cloning of a wheat cDNA (TaPSS1) that encodes a phosphatidylserine synthase (PSS) and provides the first strong evidence for the existence of this enzyme in a higher eukaryotic cell. The cDNA was isolated on its ability to confer increased resistance to aluminum toxicity when expressed in yeast. The sequence of the predicted protein encoded by TaPSS1 shows homology to PSS from both yeast and bacteria but is distinct from the animal PSS enzymes that catalyze base-exchange reactions. In wheat, Southern blot analysis identified the presence of a small family of genes that cross-hybridized to TaPSS1, and Northern blots showed that aluminum induced TaPSS1 expression in root apices. Expression of TaPSS1 complemented the yeast cho1 mutant that lacks PSS activity and altered the phospholipid composition of wild type yeast, with the most marked effect being increased abundance of phosphatidylserine (PS). Arabidopsis thaliana leaves overexpressing TaPSS1 showed a marked enhancement in PSS activity, which was associated with increased biosynthesis of PS at the expense of both phosphatidylinositol and phosphatidylglycerol. Unlike mammalian cells where PS accumulation is tightly regulated even when the capacity for PS biosynthesis is increased, plant cells accumulated large amounts of PS when TaPSS1 was overexpressed. High levels of TaPSS1 expression in Arabidopsis and tobacco (Nicotiana tabacum) led to the appearance of necrotic lesions on leaves, which may have resulted from the excessive accumulation of PS. The cloning of TaPSS1 now provides evidence that the yeast pathway for PS synthesis exists in some plant tissues and provides a tool for understanding the pathways of phospholipid biosynthesis and their regulation in plants.
Although phosphatidylserine (PS) 1 is a minor phospholipid component of plant membranes (1), it is likely to have roles in addition to contributing to the structure of lipid bilayers. In mammalian and yeast cells, PS is normally distributed asymmetrically across the plasma membrane with the inner surface containing most of the PS (2,3). Externalization of PS on the plasma membrane is an early indicator of apoptosis in mammalian cells (4,5). Phosphatidylserine is required for activation of protein kinase C (6), is implicated as a key element in the coagulation of blood (3), and also allows phagocytes to recognize apoptotic cells (7). By comparison, there is little information available describing the roles of PS in plants although a recent report describes the migration of PS to the outer surface of the plasma membrane being associated with apoptosis in plant cells (8).
In Saccharomyces cerevisiae, PS is synthesized by the action of phosphatidylserine synthase (PSS; CDP-diacylglycerol:L-serine o-phosphatidyltransferase, EC 2.7.8.8) that catalyzes the condensation of L-serine with CDP-diacylglycerol (9). The synthesis of PS in yeast is the first committed step in the major pathway for the de novo synthesis of phosphatidylethanolamine (PE) and phosphatidylcholine (PC; Fig. 1A), the two most abundant phospholipids of yeast membranes. In mammalian cells, by contrast, PS is synthesized by exchange of the polar head group of an existing phospholipid with L-serine (10) in a so-called base-exchange reaction (PSS-BE; Fig. 1B) whereas PC and PE are synthesized by pathways that do not involve PS (11). Although the enzymes involved in the synthesis of PS by both types of reactions share the same trivial name, phosphatidylserine synthase, the biochemistry of the reactions is different, and there is minimal amino acid sequence similarity between the two types of enzymes.
In plant cells the major pathway for the synthesis of PS is uncertain and may differ according to the type of tissue and species in question (12). A PSS-base exchange enzyme activity has been demonstrated in castor bean endosperm (13), whereas another report described a PSS activity in extracts of spinach leaves (14), but there are no subsequent reports to verify the presence of PSS in other plant species or tissues. Feeding of radiolabeled precursors to plant cells has produced equivocal results in identifying the major pathways for phospholipid biosynthesis. Some results have suggested that PC is synthesized primarily by pathways that do not require PS, but it has not been possible to completely rule out that PS contributes to PC biosynthesis (15,16), Despite the paucity of direct experimental evidence in plants, a dogma has developed that PS synthesis in higher eukaryotic cells, including plant cells, proceeds exclusively by the base-exchange pathway (11,17).
As a result of screening a wheat cDNA library for clones that conferred enhanced aluminum resistance to yeast, we isolated a cDNA encoding a PSS that catalyzes the formation of PS by * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  the condensation of CDP-diacylglycerol with L-serine. The cDNA complements a yeast mutant defective in PSS activity, and the putative translation product shows homology to PSS proteins from yeast and bacteria. Heterologous expression in plants leads to increased synthesis of PS, and high levels of expression are associated with lesions on the leaves.
DNA Manipulations, Sequence Analysis, and Preparation of cDNA Library-Nucleic acids were manipulated using standard methods (20) and sequenced by the dideoxy chain termination method (21).
The aluminum-resistant wheat line, ET3 (22), was grown in solution culture, and root apices (0 -5 mm) were collected from 5-day-old seedlings exposed to 10 M aluminum for times ranging from 1 to 48 h. Total RNA was extracted from the combined root apices using a method described by Chandler et al. (23). Poly(A) ϩ mRNA was purified by passage through oligo(dT)-cellulose columns (Amersham Pharmacia Biotech), and cDNA was synthesized from this RNA template using a commercial kit (Stratagene). The resulting cDNA was ligated into the pYES2 yeast expression vector (Invitrogen), and the plasmids were used to transform Escherichia coli (strain XL1-Blue; Stratagene).
Plasmid DNA was purified from the cDNA library grown in E. coli and used to transform the InVSc2 strain of S. cerevisiae using a lithium acetate procedure (24). Transformants were selected on SC minus/ uracil medium, eluted from the plates into sterile water, and screened for aluminum resistance. About 100,000 ura ϩ yeast cells were spread onto modified LPM plates that contained 2% galactose and either 100 or 150 M aluminum added as Al 2 (SO 4 ) 3 . LPM plates have low pH (pH 3.8; required to maintain the toxic Al 3ϩ species), low magnesium concentration, and low phosphorus concentration (19); in this case agar (BBL granulated agar; Becton Dickinson) was used instead of agarose, and complete amino acids were provided as described by Rose et al. (18), and uracil was omitted to provide selection for the pYES2 vector. The plates were incubated at 30°C and aluminum-resistant colonies were selected over 3 weeks. Putative aluminum-resistant colonies were retested by streaking onto aluminum-containing plates in the presence and absence of galactose (cDNA inserts were expressed behind the GAL1 promoter in pYES2) and by introducing the DNA into E. coli and then back into the yeast strain InVSc2 to confirm their ability to confer aluminum resistance.
Northern and Southern Hybridizations-RNA from root tips of aluminum-sensitive wheat (cv. Warigal) was isolated for Northern blot analysis using procedures described by Snowden and Gardner (25). Northern blots were quantitated using a Fuji film BAS 2500 Phosphor-Imager (Fuji) with MACBAS (version 2.5E) software to normalize expression levels relative to rRNA expression as determined with the wheat rRNA probe pTA250.2 (26). Genes homologous to the predicted proteins encoded by the cDNAs were identified using the BLASTX algorithm (27).
Wheat DNA was purified by the method of Doyle and Doyle (28), and Southern hybridization at high stringency was performed as described by Sambrook et al. (20).
Manipulation of the Yeast CHO1 Gene-PCR-mediated disruption of the CHO1 gene was carried out using oligonucleotides with 41 nucleotides homologous to the CHO1 gene and 19 nucleotides homologous to the HIS1 gene. The sequence of the forward primer was CGCACCTCA AGAATTCCCACACACGGACACAGACGTTATCGGGCCTCCTCTAG-TACACTC and the reverse primer was CCCAACACCAAAGCTAGAG-TGGTTGGCATAGGCAATCCCTCGGAAAGCGCGCCTCGTTCA. The PCR product (990 base pairs) was transformed into the yeast strain FY833, and transformants able to grow in the absence of histidine were screened for an inability to grow without choline. One line (CH5) was confirmed as a CHO1 disruption by PCR and Southern hybridization. Aluminum resistance of CH5 yeast strains was determined by growth in liquid LPM medium (pH 3.5) that contained histidine, choline, and galactose (2% w/v).
The yeast CHO1 gene was amplified using the primers GCAAGCT-TGGATCCAAAATGGTTGAATCAGATGAAGATTTC and CGTCTA-GAGCGGCCGCCCAGGCATGAACAAAAACTACT and cloned behind the GAL1 promoter in the vector pYES3, a modified version of pYES2 described by Smith et al. (29).
Transgenic Plants-The TaPSS1 cDNA was ligated to plant expression vectors under the control of the cauliflower mosaic virus promoter. The Arabidopsis transformation used the pART7/pART27 vector system (30) in Agrobacterium, and transgenic plants were generated by vacuum infiltration (31). For tobacco (Nicotiana tabacum), TaPSS1 was cloned into the BamHI/XbaI site of pDH51 (32), and the resulting EcoRI fragment that contained TaPSS1 with the cauliflower mosaic virus promoter and terminator was cloned into the EcoRI site of the binary vector pPLEX201-3 (33). Tobacco was then transformed with this vector using Agrobacterium and co-cultivation of leaf explants (34). Control plants were transformed with the vectors alone. For tobacco, seed (T1) was collected from the primary transgenics (T0) and grown for analysis of phospholipids and PSS activity. For Arabidopsis, kanamycin-resistant T1 plants were selected and segregating T2 lines were analyzed. In both cases, TaPSS1 overexpressing lines could be easily identified in the segregating lines on the basis of their phenotype.
Phospholipid and PSS Assays-Yeast cultures (5 ml of LPM-galactose medium) at a starting A 600 nm of 1.0 were labeled with KH 2 32 PO 4 (0.74 MBq; Amersham Pharmacia Biotech) for 1 or 24 h. Phospholipids were extracted from yeast cells using procedures described by Homann et al. (35) and were separated by thin layer chromatography using either a single dimension with chloroform:methanol:acetic acid:water (32:4:5:1 by volume) as the solvent system or with a two-dimensional procedure where chloroform:methanol:concentrated NH 4 OH:water (66: 27:3:0.9 by volume) was used for the first dimension and chloroform: methanol:acetic acid:water (32:4:5:1 by volume) was used for the second dimension. The identity of the various phospholipids was determined by comparison to standard phospholipids (Sigma), and the thin layer chromatography plates were analyzed with a PhosphorImager system (Molecular Dynamics) to quantitate the 32 P incorporated into the various phospholipids. Yeast extracts were prepared and assayed for PSS activity according to Homann et al. (35).
Similar methods were used for the assay of phospholipids and PSS activity in Arabidopsis seedlings. Seedlings (25-day-old) grown in hydroponic culture using previously described methods (36) were transferred to nutrient solution that contained 10 M phosphate and 32 P (0.74 MBq; Amersham Pharmacia Biotech). Tissues (approximately 100 mg) were ground in CH 3 Cl:isopropyl alcohol (6:1) with a hand-held homogenizer. Subsequent steps for phospholipid extraction were the

FIG. 1. The major biosynthetic pathways for PS in yeast (A) and mammalian cells (B).
The yeast enzymes that compete for the common substrate CDP-diacylglycerol are also shown. The enzymes catalyzing the various yeast reactions are as follows: arrow 1, CDPdiacylglycerol synthase; arrow 2, phosphatidylserine synthase (PSS); arrow 3, phosphatidylinositol synthase; arrow 4, phosphatidylglycerophosphate synthase; arrow 5, phosphatidylglycerophosphate phosphatase. In mammalian cells two phosphatidylserine synthase enzymes, with different substrate specificities, catalyze base-exchange reactions (PSS-BE) from pre-existing phospholipids to synthesize PS. same as used for the yeast. Extracts for PSS assay were prepared by grinding approximately 50 mg of tissue in 100 l of extraction buffer, centrifuged, and assayed according to Homann et al. (35).
Statistical analysis of the phospholipid composition in yeast and Arabidopsis was complicated by the high variability between replicates for the total 32 P incorporated. Therefore the statistical analyses were performed on the proportion of 32 P appearing in each phospholipid fraction. The data were transformed with the arcsine function to normalize the distribution and to minimize the variance between the means. For each phospholipid a t test was used to compare the means of the different genotypes.

RESULTS
Isolation of a Wheat cDNA That Encodes PSS-A wheat cDNA encoding PSS was isolated from a yeast expression library that was screened for cDNAs whose expression enhanced aluminum resistance. After screening over 2 million yeast transformants, eight cDNAs encoding six different genes were found to confer aluminum resistance to the yeast strain In-VSc2. Sequencing of these cDNAs identified one (TaPSS1) whose predicted translation product showed sequence similarity to PSS from other organisms. Fig. 2 shows that the predicted protein encoded by the TaPSS1 open reading frame is similar to PSS from S. cerevisiae (54% identity; GenBank TM accession number D00171), Bacillus subtilis (23% identity; GenBank TM accession number P39823), and Helicobacter pylori (22% identity; GenBank TM accession number AE000614). TaPSS1 also showed 20 -23% amino acid identity to putative PSS proteins predicted from sequences present in the genomes of Methanococcus jannaschii (GenBank TM accession number U67562), Mycobacterium tuberculosis (GenBank TM accession number Z84724), and Chlamydia trachomatis (GenBank TM accession number AE001355). The predicted wheat protein has eight hydrophobic regions suggesting that it is a membranespanning protein. Unlike yeast PSS, the predicted TaPSS1 protein lacks an acidic N terminus sequence (Fig. 2). The presence of a stop codon 27 bases upstream and in frame with the predicted translation start site of the TaPSS1 cDNA indicates that there is not an alternative translation start site that could account for the absence of the acidic N-terminal sequence and also suggests that the cDNA is full length.
Northern blot analysis showed that TaPSS1 mRNA was present at a low level in both wheat roots and shoots and that its expression in roots increased in response to aluminum toxicity stress (Fig. 3A). Control plants where roots were not exposed to aluminum for equivalent times did not show any change in TaPSS1  depending on the enzyme used) in the genome of hexaploid wheat (Fig. 3B).
TaPSS1 Complements the Yeast cho1 Mutant-The similarity of the protein encoded by the TaPSS1 open reading frame to PSS from yeast and bacteria suggested that it encodes a plant PSS. Yeast cho1 mutants are defective in PSS activity and require exogenously supplied choline or ethanolamine for growth (37,38). A cho1 mutant was constructed by PCR-mediated disruption in the yeast strain FY833. Fig. 4 shows that expression of the wheat TaPSS1 gene complemented the choline requirement of the cho1 mutant, demonstrating that TaPSS1 encodes a functional plant PSS enzyme.
Overexpression of this cDNA conferred aluminum resistance to another yeast strain (DBY747-a1) but did not confer aluminum resistance to strain FY833. Since overexpression of a plant PSS conferred aluminum resistance to yeast, we also determined whether altering the expression of the endogenous yeast CHO1 gene, which encodes PSS, also altered aluminum resistance. The cho1 disruption mutant in the FY833 genetic background was indeed more sensitive to aluminum than the corresponding wild type strain (data not shown). Conversely, overexpression of yeast CHO1 conferred aluminum resistance to the INVSc2 strain but not to the FY833 strain, as observed for expression of TaPSS1 (data not shown).
Expression of TaPSS1 in Yeast Alters the Composition of Phospholipids and Confers PSS Activity to a Yeast cho1 Mu-tant- Fig. 5 shows that expression of the wheat TaPSS1 gene changed the phospholipid profile of yeast strain InVSc2. Short term 32 P-labeling experiments (Fig. 5A) showed a large increase in abundance of label into PS primarily at the expense of phosphatidylinositol (PI). Longer term 32 P-labeling experiments (Fig. 5B) also showed increased PS biosynthesis at the expense of PI, but the effects were less dramatic than in the short term experiments. This difference was expected, since short term experiments label phospholipid intermediates in the various pathways, and the 32 P in each phospholipid pool gives an indication of the rate of biosynthesis. By contrast, longer term 32 P experiments more accurately reflect the steady state phospholipid composition in yeast cells (38). In long term experiments expression of TaPSS1 increased the amount of PS and PE and decreased the amount of PI, whereas the amount of PC was unchanged (Fig. 5B). Similar results were obtained for yeast strain FY833 (data not shown), which did not show enhanced aluminum resistance with expression of TaPSS1.
Overexpression of TaPSS1 in the yeast cho1 disruptant resulted in measurable PSS activity, to levels greater than those obtained by overexpression of the yeast CHO1 gene (Table I). PSS activity was undetectable in the disruption mutant transformed with the plasmid vector alone.

Overexpression of TaPSS1 in Plants Results in Increased Accumulation of PS and Is Associated with Necrotic Lesions on
Leaves-When TaPSS1 cDNA was overexpressed in Arabidopsis and tobacco under the control of the cauliflower mosaic virus promoter, a number of transgenic plants showed changes in morphology and necrotic lesions. Arabidopsis plants showing the most severe phenotypes were stunted (Fig. 6D), and the necrotic lesions on tobacco typically appeared as spots with leaves being distorted and asymmetric (Fig. 6, B and C). Lesions and altered morphology were found in plants with the highest level of TaPSS1 transcript (data not shown).
One of the Arabidopsis lines that showed a severe phenotype, D244-24, was analyzed for its phospholipid composition by 32 P labeling. After 24 h labeling with 32 P, the proportion of radiolabel into the various classes of major leaf phospholipids showed a dramatic difference in D244-24 compared with wild type Arabidopsis plants (Fig. 7A). The overexpression of TaPSS1 resulted in an increase of 32 P incorporation into PS and a corresponding decrease of 32 P incorporation into PI, as found for yeast. In addition, there was a decrease in 32 P found in phosphatidylglycerol (PG) but an increase in labeled PC. Phosphatidylglycerol, which is not a major phospholipid of yeast cells, is a major phospholipid in leaves and is found in the lamellae of chloroplasts (1). Similar results were found for PS in roots ( 32 P incorporated into PS as a percent of total phospholipids: control, 0.56 Ϯ 0.17; D244-24, 7.95 Ϯ 2.63, means Ϯ

S.E.).
We reasoned that, as in yeast cells, longer term labeling would more accurately reflect the steady state composition of plant phospholipids. Longer term 32 P labeling (7 days) still showed a much greater 32 P incorporation into PS for the TaPSS1 transgenic lines compared with control plants (Fig.  7B). As in the 24-h experiment, label incorporated into PI and PG was reduced, although the effect was less pronounced than in the short term experiment.
Phosphatidylserine synthase activity was readily detected in crude homogenates prepared from leaves of Arabidopsis and tobacco overexpressing TaPSS1 but was either undetectable or very low for control plants that were transformed with the vector alone (Table I). The levels of TaPSS1 expression activity in different transgenic lines were related to the severity of the phenotypes described above (data not shown).

DISCUSSION
Four lines of evidence support the conclusion that the wheat TaPSS1 gene encodes a functional PSS enzyme as follows: (i) there is homology between the protein encoded by TaPSS1 and known PSS proteins from yeast and bacteria; (ii) TaPSS1 complements the yeast cho1 mutant; (iii) overexpression of TaPSS1 in plants and in a disrupted cho1 mutant of yeast yields measurable PSS activity in both types of organisms; and (iv) overexpression of TaPSS1 in yeast and plants changes the phospholipid composition in both these organisms with an increase in PS content being the most marked effect. Our cloning of a functional plant PSS gene provides the first strong evidence for the existence of PSS in higher eukaryotic cells. The role of PSS in wheat could be primarily for PS biosynthesis; or alternatively, in some plant species or tissues, it may play a key role in the biosynthesis of PC, as occurs in yeast.
Complementation of the yeast cho1 mutant provided strong evidence for TaPSS1 gene function. The fact that there is a single gene encoding PSS in the yeast genome (39) precludes the possibility that TaPSS1 is enhancing endogenous PSS in yeast by transactivation of another gene. A marked difference between yeast PSS and wheat PSS is the absence of an acidic N-terminal sequence in the wheat protein (Fig. 2). However, overexpression of a truncated form of the yeast CHO1 gene that FIG. 5. Composition of phospholipids labeled with 32 P in IN-VSc2 yeast cells expressing TaPSS1 compared with vector alone. Cells were labeled with 32 P for 1 h (A) or 24 h (B) before they were analyzed for phospholipid composition. The amount of 32 P incorporated into the major phospholipids is expressed as a percent of the total 32 P incorporated into phospholipids. The means from four independent transformants for each plasmid are shown Ϯ S.E. and statistically significant differences in proportion of phospholipids between plasmids are shown by * (p Ͻ 0.05) or **(p Ͻ 0.01). Analysis by two-dimensional thin layer chromatography showed that phosphatidyldimethylethanolamine (PDE) comprised about 5% of the 32 P incorporated into phosphatidyldimethylethanolamine (PDE) ϩ PE. with TaPSS1 Crude homogenates were prepared by disruption of cells with glass beads (yeast) or by grinding leaf tissues with a hand-held homogenizer. The homogenates were centrifuged at 500 ϫ g for 15 min, and the supernatant solutions assayed for PSS activity. lacks sequence encoding the acidic N-terminal region of the protein was also able to complement the cho1 mutant. The truncated yeast gene produced less than one-tenth the amount of active enzyme produced by the wild type gene, even though comparable amounts of protein were produced by expression of the genes (40). This lower activity was attributed to the truncated form not being correctly inserted into membranes. In our case, the wheat TaPSS1 gene yielded even greater enzyme activity than the wild type CHO1 gene when overexpressed in yeast (Table I). The significance of the lack of an acidic domain in the wheat enzyme is not known, but this does not appear to limit its activity in yeast.
The plasma membrane is a primary barrier to the entry of aluminum into cells. A change in its lipid composition could conceivably alter the resistance of the cell by excluding aluminum, by altering the activity of specific membrane proteins involved in confering aluminum resistance, or by altering the response of the yeast cell to external stimuli. Aluminum ions bind to phospholipids, and the binding affinity under acidic conditions is greatest for anionic phospholipids such as PG and PI (41). Several features of our results suggest further links between phospholipids and aluminum resistance. The wheat PSS transcript was induced in response to aluminum stress, and overexpression of PSS activity increased aluminum resist-ance in two strains of yeast, whereas deletion of the PSS gene decreased aluminum resistance in a third strain. However, the change in phospholipid composition in itself does not appear to be sufficient to confer aluminum resistant to all yeast genotypes, since expression of TaPSS1 in strain FY833 did not increase aluminum resistance despite similar changes in phospholipid composition as found for strain InVSc2. Preliminary results also indicate that overexpression of TaPSS1 in tobacco does not confer enhanced aluminum resistance.
In yeast, the major pathway for de novo PC biosynthesis is through the conversion of CDP-diacylglycerol and L-serine to PS (catalyzed by PSS) which is then converted sequentially to PE and PC by the action of PS decarboxylase and methyltransferases (9). The increased PE synthesis seen in yeast cells overexpressing PSS is consistent with this pathway since PS decarboxylase should convert some of the excess PS to PE. The observed increase in PS content at the expense of PI when TaPSS1 is expressed (Fig. 5) is also consistent with PSS and PI synthase competing for CDP-diacylglycerol as a common substrate (Fig. 1A).
In plants the major pathway for PC biosynthesis is not well defined, although much of the biochemical and physiological evidence suggest that the CDP-choline pathway predominates in some tissues (12,15). The cloning of Brassica napus and Arabidopsis cDNAs that encode CTP:phosphocholine cytidylyltransferase, the enzyme that catalyzes the rate-limiting step in the conversion of choline to PC, provides support for the existence of this pathway (42,43). Our cloning of a wheat cDNA that encodes a functional PSS protein now provides evidence for the existence of the yeast pathway for PS biosynthesis in wheat. Overexpression of TaPSS1 in Arabidopsis increased PS biosynthesis at the expense of both PI and PG, consistent with PSS, PI synthase, and PG-phosphate synthase competing for CDPdiacylglycerol as a common substrate (Fig. 1A). The rate of PC synthesis was also increased in the TaPSS1-overexpressing plants (Fig. 7, A and B), but PE synthesis did not increase. This result contrasts with that found in yeast, where PE biosynthesis increased but PC was little affected in both short and long term experiments. If the pathway to PC synthesis were the same in the transgenic plants as in yeast, then one would expect PE synthesis to be enhanced relative to PC because PE is the direct precursor of PC in this pathway. One possible explanation is that a base-exchange reaction of the type found in mammalian cells ( Fig. 1B; PSS-BEI) converts PS and free choline directly to PC without proceeding through PE. The increase in PC may be a consequence of the plant cells attempting to reduce the amount of PS accumulated in the TaPSS1 overexpressers or, alternatively, may represent an active pathway for PC biosynthesis in wild type plants that is further enhanced in the transgenic plants.
As found in other organisms, PS is likely to be an essential phospholipid for growth of plant cells. In contrast to yeast and mammalian cells where PS can comprise up to 10% of the total phospholipids, in plants PS is a minor component comprising less than 2% of the total phospholipids found in leaves of many species (1). Results from our long term 32 P-labeling experiments (Figs. 5 and 7) are consistent with these values for abundance of PS in yeast and plant cells. In mammalian cells overexpressing a PSS-base exchange enzyme, PS biosynthesis was greatly increased in vivo, but the steady state phospholipid composition was unchanged (44). This lack of effect on phospholipid composition was attributed to compensatory changes in activities of the pathways that synthesize and degrade PS to maintain overall phospholipid homeostasis. Recently, Kuge et al. (45) showed that PSS-base exchange I activity is regulated by PS concentrations in the cell and that this can in itself FIG. 7. Composition of phospholipids labeled with 32 P in Arabidopsis D244-24 expressing TaPSS1 compared with control plants. Plants were labeled with 32 P for 24 h (A) or 7 days (B) before they were analyzed for phospholipid composition. The amount of 32 P incorporated into the major phospholipids is expressed as a percent of the total 32 P incorporated into phospholipids. The means Ϯ S.E. of three plants from line D244-24 and a line with vector alone are shown, and statistically significant differences in proportion of phospholipids between lines are shown by * (p Ͻ 0.05) or **(p Ͻ 0.01). account for a large part of the PS homeostasis. By contrast, our results indicate that plant cells are unable to maintain phospholipid homeostasis when PSS activity is in excess since large changes in overall phospholipid composition were seen. We speculate that the necrotic lesions arise on plants expressing TaPSS1 at high levels because of a large increase in abundance of PS in membranes. This may have resulted in PS appearing on the outer surface of the plasma membrane by saturating the mechanisms that normally maintain phospholipid asymmetry across the membrane. The appearance of PS on the outer surface of cells is an early step in the apoptotic pathway, and in mammalian cells increased biosynthesis of PS precedes the appearance of PS on the outer surface of the plasma membrane (46). Furthermore, external application of PS to Chinese hamster ovary cells can in itself trigger apoptosis, whereas other phospholipids are not able to elicit this response (47).
In conclusion, we have isolated a wheat gene that encodes a functional PSS enzyme. Overexpression of TaPSS1 in yeast and plants alters the phospholipid composition of these organisms. Altering the expression of this gene in plants provides a new tool for defining the pathways of phospholipid biosynthesis and for understanding how these pathways are regulated. Altering the abundance of PS will also provide clues to the biological functions in plants of this minor, but potentially important, phospholipid.