Plastidic Phosphatidic Acid Phosphatases Identified in a Distinct Subfamily of Lipid Phosphate Phosphatases with Prokaryotic Origin*

Plastidic phosphatidic acid phosphatase (PAP) dephosphorylates phosphatidic acid to yield diacylglycerol, which is a precursor for galactolipids, a primary and indispensable component of photosynthetic membranes. Despite its functional importance, the molecular characteristics and phylogenetic origin of plastidic PAP were unknown because no potential homologs have been found. Here, we report the isolation and characterization of plastidic PAPs in Arabidopsis that belong to a distinct lipid phosphate phosphatase (LPP) subfamily with prokaryotic origin. Because no homolog of mammalian LPP was found in cyanobacteria, we sought an LPP ortholog in a more primitive organism, Chlorobium tepidum, and its homologs in cyanobacteria. Arabidopsis had five homologs of cyanobacterial LPP, three of which (LPPγ, LPPϵ1, and LPPϵ2) localized to chloroplasts. Complementation of yeast Δdpp1Δlpp1Δpah1 by plastidic LPPs rescued the relevant phenotype in vitro and in vivo, suggesting that they function as PAPs. Of the three LPPs, LPPγ activity best resembled the native activity. The three plastidic LPPs were differentially expressed both in green and nongreen tissues, with LPPγ expressed the highest in shoots. A knock-out mutant for LPPγ could not be obtained, although a lppϵ1lppϵ2 double knock-out showed no significant changes in lipid composition. However, lppγ homozygous mutant was isolated only under ectopic overexpression of LPPγ, suggesting that loss of LPPγ may cause lethal effect on plant viability. Thus, in Arabidopsis, there are three isoforms of plastidic PAP that belong to a distinct subfamily of LPP, and LPPγ may be the primary plastidic PAP.

Plastidic phosphatidic acid phosphatase (PAP) dephosphorylates phosphatidic acid to yield diacylglycerol, which is a precursor for galactolipids, a primary and indispensable component of photosynthetic membranes. Despite its functional importance, the molecular characteristics and phylogenetic origin of plastidic PAP were unknown because no potential homologs have been found. Here, we report the isolation and characterization of plastidic PAPs in Arabidopsis that belong to a distinct lipid phosphate phosphatase (LPP) subfamily with prokaryotic origin. Because no homolog of mammalian LPP was found in cyanobacteria, we sought an LPP ortholog in a more primitive organism, Chlorobium tepidum, and its homologs in cyanobacteria. Arabidopsis had five homologs of cyanobacterial LPP, three of which (LPP␥, LPP⑀1, and LPP⑀2) localized to chloroplasts. Complementation of yeast ⌬dpp1⌬lpp1⌬pah1 by plastidic LPPs rescued the relevant phenotype in vitro and in vivo, suggesting that they function as PAPs. Of the three LPPs, LPP␥ activity best resembled the native activity. The three plastidic LPPs were differentially expressed both in green and nongreen tissues, with LPP␥ expressed the highest in shoots. A knock-out mutant for LPP␥ could not be obtained, although a lpp⑀1lpp⑀2 double knock-out showed no significant changes in lipid composition. However, lpp␥ homozygous mutant was isolated only under ectopic overexpression of LPP␥, suggesting that loss of LPP␥ may cause lethal effect on plant viability. Thus, in Arabidopsis, there are three isoforms of plastidic PAP that belong to a distinct subfamily of LPP, and LPP␥ may be the primary plastidic PAP.
The enzyme phosphatidic acid phosphatase (EC3.1.3.4) (PAP) 3 catalyzes the dephosphorylation of phosphatidic acid (PA) to yield diacylglycerol (DAG). In membrane lipid metabolism, phosphatidic acid is synthesized by two sequential steps of acylation in the so-called Kennedy pathway (1). This pathway is common to the metabolism of three different classes of glycerolipids, namely phospholipids (phosphatidylcholine and phosphatidylethanolamine), galactolipids, and triacylglycerol, and the last step in this pathway is catalyzed by PAP.
A model plant, Arabidopsis, has two pathways for glycerolipid biosynthesis, namely prokaryotic and eukaryotic pathways localized in the plastids and endoplasmic reticulum, respectively (2). In both pathways, PAP is considered to be involved in the committed step of membrane lipid biosynthesis. Therefore, PAP may be involved in lipid metabolism both in plastids and endoplasmic reticulum. Recently, a permease-like protein named trigalactosyldiacylglycerol 1 (TGD1) was isolated and shown to be involved in the import of extraplastidic PA into plastids, suggesting that certain extraplastidic substrates may be imported for plastidic metabolism (3,4). Galactolipid biosynthesis is indispensable for photosynthesis, and the importance of PAP in this process is illustrated by the fact that the first step after the two pathways meet is catalyzed by plastidic PAP (5)(6)(7).
Although information on plant PAP is scarce, native plastidic PAP activity has been studied in detail using isolated spinach chloroplast envelope (8,9). Enzyme analysis of the purified inner or outer envelope indicates that plastidic PAP is tightly associated with the inner envelope (9). Its enzymatic features suggest two outstanding qualities that distinguish plastidic PAP from other PAPs. First, extraplastidic PAP is thought to be an acidic phosphatase (10), whereas plastidic PAP has a pH optimum in an alkaline range (8,9). Second, Mg 2ϩ inhibits plastidic PAP activity, whereas extraplastidic PAPs may variously require or be independent of Mg 2ϩ (10,11).
Like other higher plants, Arabidopsis is thought to have both soluble (PAP1) and membrane-bound (PAP2) PAP (10,11). To our knowledge, however, there has been no report regarding native PAP activity in Arabidopsis. To date, four isoforms of membrane-bound PAP, designated lipid phosphate phosphatase (LPP) 1-4, have been reported that are homologous to mammalian LPP (12,13). However, they are unlikely to be plastidic because no apparent transit peptide is predicted for chloroplast localization in any isoform using a localization prediction system such as TargetP (www.cbs.dtu.dk/services/TargetP/) and WOLFpSORT (wolfpsort.seq.cbrc.jp/). Han et al. (14) recently cloned a soluble PAP, phosphatidate phosphohydrolase, for the first time in Saccharomyces cerevisiae; phosphatidate phosphohydrolase is not homologous to membrane-bound PAP. Arabidopsis has two homologs of phosphatidate phosphohydrolase, but they do not contain predictable transit peptides or membrane-spanning domains. Therefore, they are unlikely to be candidates for the plastidic PAP that is tightly associated with the inner envelope of chloroplasts. Taken together, this evidence suggests that Arabidopsis likely has additional LPPs that function as plastidic PAPs.
In an attempt to unravel the molecular characteristics and physiological function of plastidic PAP, we newly isolated a subfamily of LPP in Arabidopsis and their ancestral ortholog in the cyanobacterium Synechosystis sp. PCC6803 (SynLPP). Our results showed that three of five isoforms in Arabidopsis (LPP␥, LPP⑀1, and LPP⑀2) were plastidic PAP. Among them, LPP␥ may be the primary PAP in plastids.

EXPERIMENTAL PROCEDURES
Plant Materials-Arabidopsis thaliana Columbia-0 ecotype was used throughout the study. For plant growth, solidified Murashige and Skoog medium was used (15). T-DNA mutant lines were obtained from the Arabidopsis Biological Resource Center (Columbus, OH).
Enzyme Activity Assay-Each transformant grown in SD medium (without His, Trp, Leu, and Ura) was collected and disrupted by vortexing with glass beads. Total membrane proteins of each sample were collected by centrifugation (15,000 ϫ g, 10 min) and suspended in assay buffer (50 mM Tris-HCl, pH 7.0, 50 mM NaCl, 5% glycerol). Protein concentration in the enzyme extracts was quantified by the method of Bensadoun and Weinstein (17) with bovine serum albumin as a standard. Protein extract (50 l) was mixed with 100 l of substrate solution and 50 l of assay buffer. The substrate solution was prepared as follows: 45 nCi of L-3-PA, 1,2-di[1-14 C]palmitoyl, (PerkinElmer Life Sciences) was dispersed together with 18 nmol of L-3-PA, 1,2-dipalmitoyl in 100 l of suspension buffer (50 mM Tris-HCl, pH 7.0, 0.1% (w/v) Triton X-100). For assays in the presence of Mg 2ϩ , 50 l of 8 mM MgCl 2 dissolved in the assay buffer was added instead of assay buffer to make the final concentration 2 mM Mg 2ϩ . The assay was initiated by vigorously vortexing the protein extract and substrate solution, and the mixture was incubated at 25°C for 1 h. The reaction was stopped by vigorously vortexing with 0.5 ml of ethyl acetate and centrifuging twice at 1,500 ϫ g for 5 min with 0.5 ml of 0.45% NaCl. The upper layer was dried and dissolved in 40 l of chloroform/methanol (2:1 by v/v) for one-dimensional TLC analyses. The product solutions were developed in petroleum ether/ ethyl ether/acetic acid (50:50:1 by v/v/v) for DAG isolation. Radioactive spots were identified by comigrating commercial lipid standards and quantitated by Image Plate (Fuji Photofilm, Tokyo, Japan) and Image Analyzer (Storm, Amersham Biosciences Bioscience, Piscataway, NJ). The same protocol was used for the activity assays of intact Arabidopsis chloroplasts.
Fractionation of Plant Tissues-Intact Arabidopsis chloroplasts were isolated essentially according to Douce and Joyard (19) with minor modifications as described by Yamaryo et al. (20). To obtain microsome and soluble fractions, the remaining supernatant in chloroplast isolation was first centrifuged at 10,000 ϫ g, and the obtained supernatant was further centrifuged at 125,000 ϫ g for 30 min. The obtained pellet here was defined as microsome fraction, and supernatant was regarded as soluble fraction.
For Western blot analyses, 200 g of proteins were separated by SDS-PAGE (12.5% acrylamide) and transferred to a PROTORAN nitrocellulose membrane (Schleicher & Schuell, Dassel, Germany). Each LPP was detected with the respective antibody at a 1:500 dilution using a secondary antibody coupled to alkaline phosphatase. Preimmune sera gave no signal. To detect markers for chloroplast, microsome, and soluble fractions, antibodies (1:1,000 dilution) for light-harvesting chlorophyll-binding protein, plasma membrane aquaporin (21), and NADPH-dependent thioredoxin reductase A 4 were used.
Lipid Analysis-Total lipids were extracted by the method of Bligh and Dyer (22). Quantification of each lipid class and their fatty acid composition was conducted as described by Nakamura et al. (23).

Production of LPP␥-overexpressing Transformants and Complementation of lpp␥/lpp␥ Homozygous Mutant-To
overexpress LPP␥ in planta, full-length LPP␥ fragment was obtained from pDO105-LPP␥ vector by NotI/PstI digestion and cloned into NotI/PstI sites of pZErO-2 vector. Then the cloned LPP␥ fragment was excised by XbaI/BamHI to ligate it to the XbaI/BamHI sites of pBI121 vector. The constructed vector (pBI121-LPP␥) was introduced into WT Arabidopsis via the Agrobacterium-mediated method. Transformed plants were selected by kanamycin, and expression levels of LPP␥ were measured by Northern blotting (5 g of RNA used). One of the transformants with the highest overexpression level of LPP␥, designated LPP␥OX#5, was crossed to LPP␥/lpp␥ heterozygous mutant, and F2 generation was used for complementation assessment.
Northern Blot Analyses-Northern blot analyses were done as described previously (23).
In Vitro Pollen Tube Germination-Pollen tube germination of LPP␥/lpp␥ heterozygous mutant and WT was conducted as described previously (24).

A Strategy to Identify Plastidic PAPs in Arabidopsis-To
search for plastidic LPP isoforms in Arabidopsis, a clue was obtained from cyanobacteria. Synechocystis sp. PCC6803 has no mammalian LPP homolog, although it has membrane-bound PAP activity in vivo (data not shown). Because cyanobacteria are a potential ancestor of chloroplasts, it is possible that cyanobacteria have a distinct type of LPP that might have been inherited in chloroplasts of higher plants. Based on this hypothesis, we first tried to identify a cyanobacterial PAP. Because a direct BLAST search using known LPP sequences was unsuccessful, we identified an ancestral LPP in a more primitive organism, Chlorobium tepidum, for which the genome has been sequenced. Using a Chlorobium LPP sequence as the BLAST query, we found one homolog, sll0545, in Synechocystis sp. PCC6803 (SynLPP). Interestingly, when Arabidopsis homologs of this SynLPP were sought for, the known Arabidopsis LPP1-4 isoforms were not detected, but five LPP candidates were newly found (Table 1). These candidate genes were named LPP␤ (At4g22550), LPP␥ (At5g03080), LPP␦ (At3g58490), LPP⑀1 (At3g50920), and LPP⑀2 (At5g66450). We gave the known LPP isoforms (LPP1-4) tentative names, LPP␣1-␣4. These putative protein products have a molecular mass of 23-31 kDa, except for LPP␦, which is 46.1 kDa. Apart from LPP⑀1 and LPP⑀2, which share extremely high amino acid identity (94%), the sequence identity of these proteins varies from 28 to 47%.
The Newly Identified LPPs in Arabidopsis and Synechocystis Belong to a Distinct LPP Subgroup-The active sites of LPPs contain three conserved domains in which certain essential residues have been identified by site-directed mutagenesis (25,26). These domains are highly conserved among the newly identified LPP candidates in Synechocystis and Arabidopsis (Fig. 1A), although other regions show variability (data not shown).
We constructed a phylogenic tree using the protein sequences of the three conserved domains (Fig. 1B). The known Arabidopsis LPP␣1-␣4 isoforms were indeed close to mammalian and yeast LPP. By contrast, the cyanobacterial LPP candidates were distant from mammalian and yeast LPP but much closer to the newly identified Arabidopsis LPP candidates. These data suggest that these LPP candidates may be prokaryotic type LPPs that are distinct from the known LPP␣1-␣4, which are categorized as the eukaryotic type. The predicted amino acid sequences of these LPP candidates showed that three of them, LPP␥, LPP⑀1, and LPP⑀2, had putative transit peptides (Table 1). Because plastidic PAP activity was detected exclusively at inner envelopes, the relevant protein(s) may have transit peptide(s) targeted to chloroplasts. In this view, further studies were done for LPP␥, LPP⑀1, and LPP⑀2 as potential candidates for plastidic PAP.
Three Newly Identified Arabidopsis LPPs Localize to Chloroplasts-To assess whether the three Arabidopsis LPPs with putative transit peptides localize to chloroplasts, we raised specific antibodies and analyzed their localization by Western blotting. Because LPP⑀1 and LPP⑀2 share extremely high amino acid identities, antigens for LPP⑀1 and LPP⑀2 were synthetically prepared for specific 10-amino acid sequences (LPP⑀1, RKDLVTGGGI; LPP⑀2, RDGEDRFQAL), respectively. For LPP␥ antigen, C-terminal 14 amino acid residues of LPP␥ were employed (ARAARKDMDSAKSD). WT Arabidopsis plants were fractionated as intact chloroplasts (19,20), microsome, and soluble fraction (23) and subjected to Western blotting. As can be seen in Fig. 2, all three candidates were enriched in the purified chloroplast fraction, although the overall protein level of each isoform was so low as to be undetectable in crude extract, microsome, and soluble fractions. These antibodies 4 K. Motohashi and T. Hisabori, unpublished observations. gave single band in chloroplast fraction (supplemental Fig. S1). These results suggest that LPP␥, LPP⑀1, and LPP⑀2 were all localized mainly to the chloroplasts. The Three Plastidic LPPs Are Differentially Expressed in Green and Nongreen Tissues-We next analyzed whether the three plastidic LPPs differed in their expression levels. Semiquantitative reverse transcription-PCR using cDNA prepared from shoots showed that LPP␥ expression was predominant, whereas expression of LPP⑀1 and LPP⑀2 was low (Fig. 3A), in agreement with the result of the Massively Parallel Signature Sequencing data base (www.mpss.udel.edu/at/GeneAnalysis. php/). This suggests that LPP␥ is the most predominantly expressed isoform of plastidic PAP in shoots.
To analyze the spatial expression pattern of these three LPPs, histochemical staining with GUS was carried out using promoter::GUS fusion transformants. The three LPP isoforms were all expressed in leaves. However, there was a difference in the staining pattern in that pLPP␥::GUS was stained mainly in vascular tissues, whereas pLPP⑀1::GUS and pLPP⑀2::GUS were stained at meristematic ends of young leaves and cotyledons (Fig. 3, B-G). Furthermore, a differential GUS staining pattern was also observed in nonphotosynthetic organs; LPP␥, but not LPP⑀1 or LPP⑀2, showed strong GUS staining in flowers (Fig. 3,  H-P). Detailed analysis showed that the staining pattern of LPP␥ changed dramatically during flower development. In the juvenile buds, the whole carpel was strongly stained (Fig. 3H). This staining was then consolidated to the pistils as buds matured (Fig. 3, I and J). When filament elongation occurred, dense staining in whole anthers suddenly emerged (Fig. 3K). This staining then moved to filaments as female gametophytes became fertile (Fig. 3, L-N). In the roots, the three LPPs were strongly expressed at root tips and branch points (Fig. 3, Q-S). Thus, although the three isoforms of LPP were cooperatively expressed in leaves and roots, only LPP␥ was expressed strongly in floral organs. These results suggest that the function of LPP␥ may be nonredundant to the other two isoforms.
Plastidic LPPs and Synechocystis LPP Encode Functional PAPs-Because LPPs are all membrane-integrated proteins, these candidates were expressed in a yeast mutant line  ⌬dpp1⌬lpp1⌬pah1, which has significantly lower total PAP activity (14). When the plastidic LPP candidates as well as that of Synechocystis were introduced by transformation, membrane-bound PAP activity of the mutant yeast recovered to the wild type level, indicating that these LPPs indeed have PAP activity in vitro (Fig. 4A). Because the yeast mutant has a temperature-sensitive phenotype so that it cannot grow at 37°C (14), we evaluated whether the temperature sensitivity could be rescued by the introduction of these LPPs. Transformation of SynLPP, LPP␥, and LPP⑀2 rescued, at least partially, the mutant yeast phenotype, suggesting that these candidates also have PAP activity in vivo (Fig. 4B).
Next, the enzymatic features of these LPPs were characterized with respect to pH optimum and Mg 2ϩ dependence. Enzyme analysis of membrane proteins isolated from transformed yeast indicated that the optimal pH was 7.0 for LPP⑀1 and LPP⑀2, but LPP␥ showed a broader pH optimum, from ϳ6.0 to 8.0 (Fig. 5). These enzymes were all inhibited by the addition of Mg 2ϩ , which is a feature unique to chloroplastic PAP activity (8,9). Because Arabidopsis chloroplastic PAP activity has not been described previously, chloroplast membranes were isolated from Arabidopsis and characterized. As shown in Fig. 5, the PAP activity in chloroplast membranes showed a broad pH optimum from 6.0 to 8.0, and the activity was inhibited by the addition of Mg 2ϩ , in agreement with previous reports using spinach chloroplasts. Furthermore, when these enzymatic features were compared with those of the three prokaryotic LPP isoforms, they closely resembled that of LPP␥. These results suggest that LPP␥ may be the primary PAP in chloroplasts.
Mutant Analysis of Plastidic PAP Suggests That LPP␥ Is an Indispensable Enzyme-We next isolated T-DNA-tagged mutants of these LPPs (Fig. 6, A and B). The knock-out mutants of LPP⑀1 (lpp⑀1:SALK_000157) and LPP⑀2 (lpp⑀2: SALK_055964) were obtained from the Arabidopsis Biological Resource Center (Columbus, OH), and homozygous mutants were successfully isolated (Fig. 6A). In addition, a  double knock-out line (lpp⑀1lpp⑀2) was also produced (Fig.  6B) as a precaution against the possibility that these two isoforms might be functionally redundant because of their high amino acid sequence similarity. However, the result of mutant analyses showed that even the lpp⑀1lpp⑀2 double mutant revealed no observable phenotypical alteration under normal growth condition. Furthermore, lipid analyses of the mutant lines showed no significant differences in either lipid composition or fatty acid composition (data not shown). These results suggest that loss of LPP⑀1 and LPP⑀2 has no observable impact on the bulk of glycerolipid metabolism.
Regarding the knock-out line for LPP␥, three T-DNA tag lines were available in the Arabidopsis Biological Resource Center. However, only CS846388 had T-DNA insertion within the LPP␥ open reading frame (Fig. 6A). Although the heterozygous mutant was isolated, the homozygous line could not be obtained despite screening 100 F2 plants of the heterozygous mutant, suggesting that loss of LPP␥ might cause lethal effect on plant viability. Therefore, we investigated whether lpp␥/lpp␥ homozygous mutant can be isolated under the ectopic overexpression of LPP␥. First, we generated LPP␥-overexpressing transformant lines driven by the cauliflower mosaic virus 35S promoter (35S::LPP␥) and obtained the LPP␥OX#5 line, which showed more than 10 times higher expression levels than WT (Fig. 6C). The LPP␥OX#5 line showed no visible phenotype nor had any changes in lipid contents (supplemental Fig. S2B) and fatty acid composition (data not shown). Then this transformant was crossed with LPP␥/lpp␥ heterozygous plants, and F2 seeds were subjected to genotype analysis by PCR amplification. Because LPP␥ is a gene with single exon, the size of PCR product will be the same whether it is amplified from endogenous LPP␥ sequence or the ectopically introduced vector (35S::LPP␥). Therefore, we judged the genotype of LPP␥ by amplifying open reading frame ϩ 500 bp upstream region of LPP␥. In the F2 generation of the 35S::LPP␥ introduced LPP␥/lpp␥ heterozygous plants, PCR analysis identified a certain plant in which full open reading frame is amplified but not open reading frame ϩ 500 bp, indicating that this plant was lpp␥/lpp␥ homozygous mutant (Fig. 6D). This plant showed no differences in phenotypes from WT, indicating that overexpression of LPP␥ in planta can complement the lethal effect of LPP␥ knock out.
To further investigate the lethality of LPP␥ knock out, we observed female and male ganetophytes of LPP␥/lpp␥. Siliques of LPP␥/lpp␥ had no significant amount of aborted ovules or wrinkled seeds, suggesting that lpp␥/lpp␥ is unlikely to be embryonic lethal. On the other hand, in vitro pollen tube germination experiments showed that pollen from LPP␥/lpp␥ had a significantly lower germination rate compared with that of WT (WT, 100/301; LPP␥/lpp␥, 24/297). These results indicate that loss of LPP affects normal pollen tube germination.   lpp⑀1 and lpp⑀2, respectively. ACT8, actin 8 used for control. C, expression levels of LPP␥ in WT and LPP␥OX#5, a LPP␥ overexpressing transformant in WT background. 5-g of total RNA were used for Northern blotting. D, genomic PCR for WT, LPP␥/lpp␥ heterozygous mutant, and lpp␥/lpp␥ harboring LPP␥OX#5 transformation. The primers used were: Fw, a forward primer starts from the sequence encoding start codon; Ϫ500Fw, a forward primer starts from 500 bp upstream of the sequence encoding start codon; Rv, a reverse primer starts from the sequence encoding stop codon.
An independent experiment was designed to generate RNAisuppressed LPP␥ knock-out/knock-down lines because complete knock-out of LPP␥ caused lethal effect. We isolated 30 independent transformants harboring LPP␥-RNAi. The expression level of LPP␥ was decreased to some extent in most of them, with LPP␥-RNAi#5 and #8 showing the least expression level (ϳ15% of WT; supplemental Fig. S2A). However, these lines showed no impact on lipid composition (supplemental Fig. S2B), suggesting that a partial suppression (ϳ15% of WT) may be dispensable for plant viability. Considering that we could not obtain any transformants that showed further decrease in LPP␥ expression, those with a strongly suppressed line (below 15%) might become lethal and therefore unable to be obtained by screening. Thus, the results of mutant analyses suggest that LPP␥ is an indispensable PAP for plant viability, whereas LPP⑀1 and LPP⑀2 are unlikely to contribute significantly to the lipid metabolism, at least under normal growth conditions.

DISCUSSION
The phylogenetic tree constructed by comparison of the three conserved domains of the LPP family (LPP motifs) showed that LPPs are clearly divided into two subfamilies: either LPP isoforms in human, yeast, and Arabidopsis or those in cyanobacteria and their homologs in Arabidopsis. This classification suggests that the LPP isoforms may be categorized either as eukaryotic or prokaryotic LPPs, respectively. The fact that prokaryotic PAP is detected only in bacteria and higher plants leads us to presume that prokaryotic PAP might have been introduced into higher plants by endosymbiosis, thereby yielding plastidic PAP. Because the prokaryotic PAPs found in Arabidopsis and cyanobacteria have very low amino acid similarity to known eukaryotic LPPs, they were undetectable by a conventional homology search approach. However, three LPP domains are highly conserved in all Arabidopsis prokaryotic LPPs. Site-directed mutagenesis of these domains in yeast or mammalian LPP suggests that these domains are critical for enzymatic activity (25,26), and together they may form an active site because they all face the same side of the membrane in every case studied so far (27). However, using a deduced membrane topology of these LPPs that consider membrane-spanning regions, domain 3 of prokaryotic LPP in Arabidopsis faces the opposite side (predicted by SOSUI program, available at bp.nuap.nagoya-u.ac.jp/ sosui/). Considering that these LPPs also exhibited significant PAP activity both in vitro and in vivo, it is likely that domain 3 is not a part of the active site but might have an alternate function. This idea is also supported by the evidence that SynLPP showed significant PAP activity in vitro and perfectly complemented the temperature-sensitive phenotype of mutant yeast in vivo, even though it partially lacks the highly conserved residues (Ser and Arg) in domain 3. Therefore, it is of enzymological interest to determine the functions of each domain by comparatively analyzing both eukaryotic and prokaryotic LPP families.
The three prokaryotic LPPs in Arabidopsis, LPP␥, LPP⑀1, and LPP⑀2, all localized to the chloroplast. In each protein, membrane-spanning regions were identified by the SOSUI program, suggesting that these enzymes are integral membrane proteins. This is in good agreement with a previous report on spinach chloroplasts indicating that chloroplastic PAP activity is tightly associated with the inner envelope (9). Considering that all three LPPs have apparent transit peptide sequences and that the previous report detected no significant PAP activity in purified outer envelope (9), these LPPs are likely to be integrated in the inner envelope of chloroplasts. As for the topological orientation of these LPPs, it seems consistent that PAP faces the stromal side of the inner envelope because glycerol-3-phosphate acyltransferase is a stromal protein (28) and lysophosphatidic acid acyltransferase activity faces the stromal side of the inner envelope (29). In addition, a recent report on TGD1 (4) suggests that it might deliver PA as far as the stromal side of the inner envelope. However, monogalactosyldiacylglycerol synthase 1, an enzyme that acts subsequent to PAP, faces the intermembrane space, raising the question of how the substrate DAG penetrates the inner envelope from the stromal side to the opposite side (4). The reaction steps leading from PA to monogalactosyldiacylglycerol are rather rapid and thought to be highly regulated because chloroplastic PAP activity is negatively regulated by the product DAG (30), and PA induces monogalactosyldiacylglycerol synthase activity (31). Therefore, it is important to elucidate the membrane topology of the chloroplastic LPPs. However, to do so, one must first determine which regions of LPPs are involved in substrate binding or catalytic activity, because they are integral membrane proteins that have several exposed regions on each side of the membrane. Because the conventional idea that the three LPP domains form an active site is not applicable for plastidic LPPs, future studies are required to identify the essential regions of the enzyme and determine its topology.
The comparison of expression levels among the three plastidic LPPs revealed that LPP␥ expression was highest in shoots. This result was supported by the data distributed by the Massively Parallel Signature Sequencing data base that LPP␥ is expressed at more than 10 times higher levels than the other isoforms. Histochemical GUS staining showed that all these LPPs were expressed in leaves. However, there was a difference in the staining pattern in that LPP␥ was expressed in vascular tissues, whereas LPP⑀1 and LPP⑀2 were expressed at the meristematic end of young leaves and cotyledons. On the other hand, GUS staining was detected in flowers only for LPP␥, which showed a dynamic GUS staining pattern during flower development. Although there are plenty of examples showing specific expression in flowers, the distinctiveness of the LPP␥ pattern is that it was expressed temporally and spatially in both male and female gametophytes. These results suggest that LPP␥ may be a predominant plastidic PAP in photosynthetic organs and play a unique role in development of floral organs. The lethal effect by LPP␥ knock-out might reflect partly the functional importance of this isoform in floral organs.
The expression of recombinant LPP in the yeast mutant ⌬dpp1⌬lpp1⌬pah1 showed significant recovery of total PAP activity for the four LPPs analyzed. Furthermore, the tempera-ture-sensitive phenotype of the mutant yeast was, at least in part, recovered by the introduction of some of these LPPs. These results suggest that the newly identified LPPs function as PAPs both in vitro and in vivo. Enzymatic characterization of these LPPs showed that their PAP activities were inhibited by Mg 2ϩ . The effective inhibitory concentration of Mg 2ϩ was consistent with that reported in isolated spinach chloroplasts (8,9). However, this is in contrast to data for the eukaryotic LPPs in Arabidopsis (LPP␣1 and ␣2) in that Mg 2ϩ does not inhibit PAP activity and even activates it (12). This suggests that the enzymatic properties of plastidic LPPs are distinct from known Arabidopsis LPPs but similar to those of intact chloroplast membranes isolated from spinach and Arabidopsis. Furthermore, the pH profiles for the PAP activity of LPP⑀1 and LPP⑀2 had a sharp peak at pH 7, whereas that for LPP␥ showed a broader maximal activity from pH 6 to 8, which corresponds well to that of isolated Arabidopsis chloroplasts. The fact that the inhibitory effect by Mg 2ϩ and optimal pH were also similar between LPP␥ and Arabidopsis chloroplasts again suggests that LPP␥ activity predominates among the three chloroplastic LPPs. Although analysis with the isolated spinach chloroplast envelope characterized chloroplastic PAP as an alkaline phosphatase (8,9), this might not be applicable to other species because chloroplastic PAP in Arabidopsis showed a broader pH optimum from slightly acidic to weakly alkaline under our assay condition.
To study the function of the three plastidic LPPs in vivo, T-DNA tagged mutants were isolated. Although homozygous mutants of LPP⑀1 and LPP⑀2 were isolated successfully and a double knock-out mutant was also produced, no significant changes in lipid composition or fatty acid composition of each lipid class were observed even in the double knock-out mutant. By contrast, the homozygous mutant of LPP␥ was not obtained even by self-crossing the heterozygous mutant (n ϭ 100). However, the lpp␥/lpp␥ homozygous mutant was isolated only under ectopic overexpression of LPP␥ in planta, suggesting that LPP␥ is an indispensable enzyme for plant viability. Although it is currently unknown why the loss of LPP␥ is lethal, one possible interpretation is that LPP␥ plays a crucial role in lipid metabolism. Because suppression of TGD1 severely affects the eukaryotic pathway (3,4), a significant portion of eukaryotic lipid may be imported into chloroplasts in a form of PA through TGD1. In addition, crossing of tgd1 with act1, in which prokaryotic lipid metabolism is abolished (32), resulted in embryonic lethal phenotype (4). As mentioned in the introduction, plastidic PAP is considered to be involved in both eukaryotic and prokaryotic pathways because PA is the major extraplastidic lipid to be incorporated into plastids (4). Therefore, it is expected that knock-out of plastidic PAP also shows lethality as was observed in act1tgd1 (4). The reason lpp␥/lpp␥ was not embryonic lethal but showed possible defects in pollen elongation might be that in addition to the primary involvement in leaf PAP activity, LPP␥ plays crucial roles in a certain process during male gametophyte development. The strong GUS staining in floral organs and results of in vitro pollen tube germination experiment support this idea. Because lpp␥/lpp␥ homozygous mutant is not available and RNAi knock-down strategy was unsuccessful so far, the remaining question is how LPP␥ is involved in lipid metabolism in leaves as well as flowers. In this regard, it may be useful in future study to selectively express individual genes in specific tissues using the gene induction systems such as the GVG system (33) or the XVE system (34). Any phenotypical changes in leaves or floral organs may extend our understanding of the function of LPP␥ in vivo.
In summary, we isolated one cyanobacterial PAP in Synechocystis sp. PCC6803 (SynLPP) and three homologous plastidic PAPs in Arabidopsis (AtLPP␥, LPP⑀1, and LPP⑀2) that belong to a distinct subfamily (prokaryotic type) of LPPs. Characterization of the three plastidic PAP isoforms suggests their differentiated function, with which LPP␥ may predominate over LPP⑀1 and LPP⑀2. Further studies are expected to uncover the full scope of PAP in Arabidopsis.