Phospholipase C5 (NPC5) is involved in galactolipid accumulation during phosphate limitation in leaves of Arabidopsis


  • Nicole Gaude,

    1. Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Potsdam, Germany
    Search for more papers by this author
  • Yuki Nakamura,

    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, 4259-B-65 Nagatsuta-cho, Midori-ku, Yokohama 226-8501, Japan
    Search for more papers by this author
    • Present address: Temasek Life Sciences Laboratory, National University of Singapore, Singapore 117604, Singapore.

  • Wolf-Rüdiger Scheible,

    1. Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Potsdam, Germany
    Search for more papers by this author
  • Hiroyuki Ohta,

    1. Center for Biological Resources and Informatics and Research Center for the Evolving Earth and Planets, Tokyo Institute of Technology, 4259-B-65 Nagatsuta-cho, Midori-ku, Yokohama 226-8501, Japan
    Search for more papers by this author
  • Peter Dörmann

    Corresponding author
    1. Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Potsdam, Germany
    Search for more papers by this author

*(fax +49 331 567 8250; e-mail


The replacement of phospholipids by galacto- and sulfolipids in plant membranes represents an important adaptive process for growth on phosphate-limiting soils. Gene expression and lipid analyses revealed that the MYB transcription factor PHR1 that has been previously shown to regulate phosphate responses is not a major factor controlling membrane lipid changes. Candidate genes for phospholipid degradation were selected based on induction of expression during phosphate deprivation. Lipid measurements in the corresponding Arabidopsis mutants revealed that the non-specific phospholipase C5 (NPC5) is required for normal accumulation of digalactosyldiacylglycerol (DGDG) during phosphate limitation in leaves. The growth and DGDG content of a double mutant npc5 pho1 (between npc5 and the phosphate-deficient pho1 mutant) are reduced compared to parental lines. The amount of DGDG increases from approximately 15 mol% to 22 mol% in npc5, compared to 28 mol% in wild-type, indicating that NPC5 is responsible for approximately 50% of the DGDG synthesized during phosphate limitation in leaves. Expression in Escherichia coli revealed that NPC5 shows phospholipase C activity on phosphatidylcholine and phosphatidylethanolamine. A double mutant of npc5 and pldζ2 (carrying a mutation in the phospholipase Dζ2 gene) was generated. Lipid measurements in npc5 pldζ2 indicated that the contribution of PLDζ2 to DGDG production in leaves is negligible. In contrast to the chloroplast envelope localization of galactolipid synthesis enzymes, NPC5 localizes to the cytosol, suggesting that, during phosphate limitation, soluble NPC5 associates with membranes where it contributes to the conversion of phospholipids to diacylglycerol, the substrate for galactolipid synthesis.


Phosphate is one of the most important macronutrients, and deficiency of phosphate limits crop production in many parts of the world (Raghothama, 1999, 2000). Phosphate is taken up by the roots and transported to the shoot via the xylem (Daram et al., 1999; Versaw and Harrison, 2002). While large amounts of inorganic phosphate are stored in the vacuole, a proportion is incorporated into organic compounds, i.e. nucleic acids (RNA and DNA), phosphorylated proteins, nucleotides, sugar phosphates and phospholipids (Poirier et al., 1991). The molecular basis for the sensing and regulation of phosphate deficiency is only partially understood. Two Arabidopsis mutants, pho1 and pho2, which respectively show reduced or increased amounts of phosphate in leaves relative to wild-type (WT), have been isolated (Delhaize and Randall, 1995; Poirier et al., 1991). The PHO1 gene encodes a transporter involved in phosphate allocation from the root epidermis and cortex to the xylem (Hamburger et al., 2002). The microRNA miR399 has been shown to control phosphate homeostasis by regulating expression of a component of the ubiquitin-dependent protein degradation pathway (Chiou et al., 2006). This component, an E2-conjugating enzyme, was identified as the PHO2 gene of Arabidopsis (Aung et al., 2006; Bari et al., 2006). Furthermore, PHR1 (phosphate response 1), was identified as a MYB transcription factor that regulates some responses to phosphate deprivation, including anthocyanin accumulation and proliferation of the root system (Rubio et al., 2001). PHR1 controls transcription of the microRNA miR399, thereby regulating PHO2 expression. However, PHR1 is not involved in all phosphate responses, and alternative signaling mechanisms must exist in plants (Bari et al., 2006; Stefanovic et al., 2007).

Various adaptive mechanisms have evolved in higher plants to cope with phosphate limitation. During phosphate deprivation, expression of several genes is induced, resulting in an increase in phosphate uptake and transport capacity (Mudge et al., 2002; Ticconi and Abel, 2004). Induction of RNase gene expression stimulates RNA degradation, thereby releasing phosphate for other cellular processes (Bariola et al., 1994; Taylor et al., 1993). The alteration of membrane lipid composition represents another mechanism that is stimulated during phosphate deprivation. In phosphate-deprived plants, the amounts of sulfolipid (sulfoquinovosyldiacylglycerol, SQDG) and of the galactolipid digalactosyldiacylglycerol (DGDG) increase at the expense of phospholipids (Dörmann and Benning, 2002;Essigmann et al., 1998; Härtel et al., 2000). DGDG serves as a surrogate for phospholipids in plastidial and extra-plastidial membranes (Andersson et al., 2003; Härtel et al., 2000; Jouhet et al., 2004). The change in membrane lipids is accompanied by upregulation of galactolipid and sulfolipid gene expression (Awai et al., 2001; Essigmann et al., 1998; Kelly and Dörmann, 2002; Kelly et al., 2003; Yu et al., 2002).

Phospholipases have been suggested to be involved in phospholipid degradation, thereby producing phosphatidic acid and diacylglycerol for galactolipid and SQDG synthesis. Based on radioactive labeling studies, a phospholipase D was found to be critical for phospholipid degradation (Andersson et al., 2004). The phospholipases PLDζ1 and PLDζ2 have been shown to be involved in DGDG accumulation during phosphate deprivation in the root, but not in the leaf (Cruz-Ramírez et al., 2006; Li et al., 2006a,b). In addition, a gene family of non-specific phospholipases C (NPC), presumably involved in phosphate-dependent lipid changes, has been described (Nakamura et al., 2005). Expression of one of these genes, NPC4, was strongly upregulated during phosphate deprivation. However, lipid measurements in the npc4 mutant provided no evidence for an in vivo role of NPC4 in phospholipid degradation and DGDG synthesis. To study the function of phospholipid-degrading enzymes in a comprehensive manner, we have analyzed the expression of candidate genes and lipid changes in Arabidopsis mutants. From these studies, it became clear that PHR1 is not the main factor controlling membrane lipid changes, and that the non-specific phospholipase C5 (NPC5) is critical for phospholipid degradation and DGDG accumulation in leaves of Arabidopsis.


Accumulation of the galactolipid DGDG during phosphate deprivation is independent of PHR1

To determine whether the increase in galactolipid synthesis during phosphate deprivation depends on PHR1, a T-DNA mutant allele of phr1 was obtained. Changes in galactolipid gene expression were recorded by Northern blot in WT, phr1 and the pho1 mutant (Figure 1a). Due to a defect in xylem loading, leaves of pho1 are permanently phosphate-deprived, and pho1 therefore represents an alternative, genetic model to study phosphate limitation (Hamburger et al., 2002; Härtel et al., 2000; Poirier et al., 1991; Stefanovic et al., 2007). Expression of the galactolipid synthases MGD2, MGD3, DGD1 and DGD2 was induced in pho1 and in WT plants raised on phosphate-deficient medium (Figure 1). PHR1 expression was abolished in phr1, indicating that it carries a null allele. Expression of the galactolipid genes MGD2, MGD3, DGD1 and DGD2 was still upregulated in phr1 during phosphate limitation, but the extent of induction was attenuated for MGD2, MGD3 and DGD1. Quantification of membrane lipids revealed that DGDG accumulation was not compromised in phr1 (Figure 1c). Therefore, although the phr1 mutation exerts a limited control on induction of galactolipid gene expression, the accumulation of DGDG during phosphate limitation is not affected.

Figure 1.

 Galactolipid synthesis during phosphate deprivation is independent of PHR1.
(a) Location of the T-DNA insertion in the phr1 mutant allele of Arabidopsis. Exons are indicated by boxes.
(b) Gene expression was recorded by Northern blot analysis. WT and pho1 plants were grown on soil, and WT and phr1 mutants were raised on agar medium in the presence or absence of phosphate as indicated. The top panels indicate the Northern signals after hybridization to PHR1, MGD2, MGD3, DGD1 or DGD2. The bottom panel represents the ethidium bromide-stained 25S rRNA band of the gel before blotting.
(c) DGDG content in leaves of WT and phr1 plants raised on phosphate-containing or phosphate-free agar medium was measured by TLC/GC of fatty acid methyl esters (n = 3, mean and SD).

Expression of genes involved in phospholipid degradation

The conversion of phospholipids to galactolipid requires the cleavage of phospholipids by phospholipases or other phospholipid-degrading enzymes, and subsequent conversion of phosphatidic acid and diacylglycerol to monogalactosyldiacylglycerol (MGDG) and DGDG. Genes of phospholipid metabolism induced during phosphate deprivation were identified by expression profiling experiments using Arabidopsis oligonucleotide arrays (Misson et al., 2005;Morcuende et al., 2007; Wu et al., 2003). Expression of some genes was upregulated in shoots and roots of Arabidopsis seedlings, i.e. phospholipase Dζ2 (PLDζ2, At3g05630), glycerophosphodiester phosphodiesterase 1 (GPD1, At5g08030), and two related non-specific phospholipases C (NPC4, At3g03530; NPC5, At3g03540) (Figure 2). Arabidopsis mutants containing T-DNA insertions between the start and stop codons of the genes PLDζ2, GDP1, NPC4 and NPC5 were obtained (Figure 3a). Northern hybridization confirmed that expression of all four genes was upregulated in the pho1 mutant or in WT plants raised without phosphate (Figure 3b). The nucleotide sequences of the NPC4 and NPC5 coding regions are highly similar (85.1% identity), and, as a consequence, expression of these two genes cannot be distinguished in transcript profiling experiments using Affymetrix DNA arrays (Figure 2). However, using 5′-UTRs, the expression of these two genes could be distinguished in Northern blots (Figure 3b).

Figure 2.

 Pathways for the conversion of phospholipids to galactolipids during phosphate deprivation.
Phospholipids can be hydrolyzed by phospholipases C or D. Alternatively, after cleavage of acyl groups by phospholipases A, the head group of the resulting glycerophosphodiester can be hydrolyzed. Glycerol-3-phosphate can be converted into phosphatidic acid (PA) by two acyltransferase reactions. PA can be converted to diacylglycerol by PA phosphatase and used for galactolipid synthesis. Genes whose expression is induced during phosphate deprivation are shaded in grey. Expression patterns (insets) are derived from Affymetrix chip hybridization experiments of plants grown in liquid culture (Morcuende et al., 2007). The bars show expression values (mean and SD of three experiments; relative values) for shoot with phosphate (grey), shoot without phosphate (white), root with phosphate (grey hatched) and root without phosphate (white hatched). Note that only one inset is shown for NPC4/NPC5 because these genes cannot be distinguished by expression analysis on Affymetrix chips (see text). Chol, choline; P, phosphate.

Figure 3.

 Expression of genes of phospholipid degradation during phosphate deprivation.
(a) T-DNA insertion mutants pldζ2, gpd1, npc5-1 (primer LBb1) and npc5-2 (primer F-RB). The positions of the T-DNAs are shown with respect to introns (lines) and exons (boxes). The 3′ part of the 11th exon of PLDζ2 is omitted. Oligonucleotides are indicated by arrows.
(b) Induction of PLDζ2, GPD1, NPC4 and NPC5 expression during phosphate deprivation. Northern blots of leaf RNA from WT and pho1 raised on soil, and from WT, pldζ2, gpd1, npc4, npc5 and phr1 raised on agar medium with (+P) or without (−P) phosphate are shown. The top and bottom panels for each gene indicate hybridization signals and 25S rRNA as a loading control, respectively.

Expression of the corresponding mutant genes in the lines gpd1, pldζ2, npc4 and npc5-1 was abolished. In Northern blots of pldζ2 and gpd1 leaf RNA, weaker bands with aberrant sizes were observed (Figure 3b). The insertion into the pldζ2 locus altered phosphate-dependent expression of the aberrant gene fragment as indicated by the observation that its band intensity was higher under normal than under phosphate-limited conditions. Taken together, the location of T-DNA insertions between the start and stop codons in the mutants and the results of Northern blot experiments strongly support the conclusion that the mutations in pldζ2, gpd1, npc4 and npc5 represent null alleles. Northern blot experiments were also performed using the phr1 mutant (Figure 3b). Expression of PLDζ2, GDP1 and NPC5 was upregulated in phr1 upon phosphate deprivation, but the expression level was compromised compared to wild-type, and was strongly suppressed for NPC4. Therefore, similar to the genes of galactolipid synthesis, upregulation of PLDζ2, GDP1 and NPC5 expression during phosphate limitation was partially attenuated in the phr1 mutant background.

NPC5 is critical for growth and DGDG synthesis during phosphate deprivation

To address the question of whether the genes PLDζ2, GPD1, NPC4 or NPC5 affect DGDG accumulation, the corresponding mutants were exposed to phosphate-limiting conditions, and membrane lipid composition was measured. In WT leaves, the amount of DGDG increases from 15 mol% to approximately 28 mol% during phosphate deprivation (Figure 4) (Härtel et al., 2000). A similar increase in DGDG was found for leaves of pldζ2, gpd1 and npc4, indicating that the respective gene products do not contribute to the shift in lipid composition. However, in npc5-1 leaves, DGDG accumulation during phosphate deprivation was impaired, with DGDG increasing to only 22 mol% (Figure 4). Analysis of a second independent mutant allele, npc5-2, confirmed this result.

Figure 4.

 DGDG accumulation during phosphate deprivation is compromised in leaves of npc5.
Amounts of membrane lipids [MGDG, DGDG, SQDG, phosphatidylglycerol (PG), PE, PC] in leaves (left panels) or roots (right panels) of WT and the mutants pldζ2, gpd1, npc4, npc5-1, npc5-2 and npc5-1 pldζ2 after phosphate deprivation. Plants were raised on medium with phosphate (black bars) or without phosphate (grey bars). Leaf glycerolipids were separated by TLC and measured by GC of fatty acid methyl esters (mean and SD, = 3).

Galactolipids were also measured in root tissues. While DGDG accumulation in roots of WT and npc4 was very similar, it was slightly impaired in npc5-1, indicating that npc5-1 contributes to DGDG accumulation in roots to a minor extent (Figure 4). The fatty acid pattern of a membrane lipid is indicative of its origin from the prokaryotic/plastidial or eukaryotic/ER pathway of lipid synthesis. The extra-plastidial pool of DGDG in leaves contains high amounts of 16:0, particularly at the sn-1 position of glycerol (Härtel et al., 2000). During phosphate limitation, the 16:0 content of DGDG increases. The fact that the accumulation of 16:0 in DGDG is impaired in npc5-1 suggests that NPC5 primarily affects the eukaryotic/extra-plastidial pathway of lipid synthesis (Table 1).

Table 1.   Fatty acid composition of galactolipids in npc5-1 leaves
LipidFatty acidWTnpc5-1
  1. Plants were grown on agar medium in the presence (+P) or absence (−P) of phosphate as described in Experimental procedures. Lipids were separated by TLC, and the fatty acid composition of lipids was determined by GC of methyl esters. Data represent means ± SD of three experiments.

MGDG16:01.2 ± 0.43.3 ± 0.41.1 ± 0.1 1.1 ± 0.0
16:10.8 ± 0.00.8 ± 0.20.3 ± 0.10.3 ± 0.0
16:21.3 ± 0.01.4 ± 0.1 0.9 ± 0.01.0 ± 0.0
16:332.0 ± 1.627.7 ± 0.532.9 ± 0.230.2 ± 0.1
18:00.2 ± 0.00.5 ± 0.2 0.2 ± 0.00.2 ± 0.0
18:10.6 ± 0.10.9 ± 0.1 0.3 ± 0.00.3 ± 0.0
18:22.7 ± 0.34.4 ± 0.1 2.5 ± 0.22.9 ± 0.0
18:361.1 ± 0.861.0 ± 1.061.6 ± 0.264.0 ± 0.1
DGDG16:011.5 ± 1.218.9 ± 0.510.5 ± 0.6 13.4 ± 0.4
16:10.3 ± 0.10.4 ± 0.1 0.3 ± 0.20.2 ± 0.1
16:20.6 ± 0.00.4 ± 0.1 0.5 ± 0.00.4 ± 0.0
16:32.5 ± 0.11.4 ± 0.1 2.4 ± 0.01.7 ± 0.1
18:00.8 ± 0.11.3 ± 0.2 0.9 ± 0.01.0 ± 0.1
18:11.0 ± 0.21.2 ± 0.1 0.8 ± 0.10.7 ± 0.0
18:24.7 ± 0.312.8 ± 0.1 4.6 ± 0.17.9 ± 0.2
18:378.5 ± 1.563.4 ± 1.179.5 ± 1.074.6 ± 0.4

The fresh weights of 4-week-old WT, npc5-1 and npc5-2 plants grown on phosphate-sufficient medium were very similar (116 ± 27, 115 ± 26 and 103 ± 13 mg, respectively; = 10), indicating that the npc5 mutation does not affect overall plant physiology under phosphate-sufficient conditions. During phosphate deprivation, growth of WT plants was retarded but the plants were significantly larger than npc5-1 and npc5-2 plants (27 ± 8, 14 ± 4 and 15 ± 3 mg, respectively; = 19; < 0.01). Therefore, the phospholipase NPC5 is important for normal growth and DGDG production during phosphate limitation.

An independent approach to study the role of NPC5 in DGDG accumulation involved analysis of a double mutant of pho1 and npc5-1. This strategy has already been used for analysis of pho1 dgd1 plants (Härtel et al., 2000). The pho1 mutant represents a convenient model to study phosphate deprivation because the leaves of pho1 are permanently phosphate-deprived, but the physiological response of WT plants to phosphate limitation may vary dependent on the phosphate status of the leaves. The npc5-1 mutant grows normally, but growth of the double mutant pho1 npc5-1 is severely retarded compared to pho1, indicating that NPC5 is required for normal growth under phosphate limitation (Figure 5a). The inhibition of phosphate allocation to the leaves of pho1 leads to an increase in DGDG content from approximately 15 mol% (WT) to 27 mol% (Figure 5b) (Härtel et al., 2000; Poirier et al., 1991). The DGDG content of leaves of npc5-1 plants grown on soil was very similar to that of WT, but pho1 npc5-1 double mutant plants showed an intermediate DGDG content of only 22 mol%.

Figure 5.

 DGDG accumulation is compromised in the pho1 npc5-1 double mutant grown on soil.
(a) Growth of pho1 npc5-1 double homozygous plants is more severely affected compared to the parental lines pho1 and npc5-1. From left to right: WT, pho1, npc5-1 and pho1 npc5-1.
(b) Membrane lipid composition in leaves of WT, pho1, npc5-1 and pho1 npc5-1. The DGDG level is increased to 27 mol% in leaves of pho1. Introduction of the npc5-1 mutation into the pho1 mutant background attenuates the DGDG accumulation to 22 mol%. Membrane lipids were quantified by TLC and GC of fatty acid methyl esters (= 3, mean and SD).

The changes in DGDG accumulation during phosphate deprivation in npc5 lines were accompanied by alterations in the amounts of other membrane lipids. The accumulation of SQDG seemed to be compromised in leaves of the two npc5 mutant alleles npc5-1 and npc5-2 (Figure 4), and was lower in pho1 npc5-1 plants compared to pho1 (Figure 5), suggesting that NPC5 might also be important for sulfolipid accumulation during phosphate deprivation. During phosphate limitation, the amounts of phospholipids [particularly phosphatidylethanolamine (PE) and phosphatidylcholine (PC)] remained higher in npc5 lines compared to WT. Furthermore, the amounts of PE and PC in npc5-1 pho1 were elevated compared to pho1, demonstrating that NPC5 exerts control over phospholipid turnover under phosphate limitation.

Figure 4 shows that NPC5 is responsible for approximately 50% of the DGDG produced during phosphate limitation in leaves. The remaining amount of diacylglycerol required for galactolipid production might be derived from the activities of alternative phospholipases. The phospholipase PLDζ2 has previously been shown to be involved in DGDG production under phosphate limitation in roots (Cruz-Ramírez et al., 2006; Li et al., 2006a,b). An npc5-1 pldζ2 double mutant was generated to measure the extent to which the two phospholipases contribute to phosphate-dependent DGDG accumulation in leaves. DGDG accumulation in npc5-1 pldζ2 at low phosphate concentration was not further diminished compared to the parental line npc5-1, indicating that DGDG production in leaves is largely independent of PLDζ2.

NPC5 encodes a phospholipase C with specificity for PE and PC

To compare the relative contributions of NPC4 and NPC5 to overall non-specific phospholipase C activity in leaves, WT and npc mutant plants were raised under phosphate-limiting conditions, and PC-hydrolyzing activity was measured. NPC activity in npc4 is strongly reduced, but the activity in the npc5-1 mutant is not affected (Figure 6a). Therefore, in accordance with previous results (Nakamura et al., 2005), NPC4 is the most active NPC during phosphate deprivation, while NPC5 shows only low activity.

Figure 6.

 Phospholipase C activity of NPC4 and NPC5.
(a) Phospholipase C activity in leaf protein extracts of the npc4 and npc5-1 mutants after phosphate deprivation. Plants were germinated on Murashige and Skoog (1962) medium. Enzyme activity was measured in leaves 2 weeks after transfer of the plants to medium lacking phosphate.
(b) Expression of NPC4 and NPC5 in recombinant E. coli. Protein was extracted from NPC4- or NPC5-expressing cells with (+) or without (−) induction by IPTG, separated by polyacrylamide gel electrophoresis and stained with Coomassie brilliant blue. Arrows indicate accumulation of the 60 kDa NPC proteins after induction.
(c) Phospholipase C activity of protein extracts from E. coli expressing NPC4 or NPC5 was measured using radioactive PC. Data represent means and SD of three measurements.
(d) Reaction products of phospholipase assays using radioactive PE or PC were separated by TLC and visualized by autoradiography. Note that E. coli harbors phospholipase A activity resulting in free fatty acid (FFA) production. Expression of NPC4 or NPC5 results in the production of diacylglycerol (DAG).

The NPC5 cDNA was heterologously expressed in Escherichia coli. After induction of expression and polyacrylamide gel electrophoresis, protein bands corresponding to the sizes of the open reading frames of NPC4 and NPC5 were identified at 60 kDa (Figure 6b). Phospholipase C activity measured with PC was 40-fold higher for NPC4 (114 ± 7 pmol min−1 mg−1) than for NPC5 (2.6 ± 1.6 pmol min−1 mg−1) (Figure 6c). Separation of reaction products by TLC revealed that the E. coli control extract contains phospholipase A activity, resulting in the production of free fatty acids. Overexpression of NPC5 led to conversion of radioactive PE or PC to a product co-migrating with diacylglycerol, clearly demonstrating that NPC5 harbors phospholipase C activity (Figure 6d).

The phospholipase NPC5 localizes to the cytosol

To determine the subcellular localization of NPC5, the cDNA was N-terminally fused to GFP and transiently expressed in Arabidopsis leaves. GFP fluorescence of NPC5–GFP and NPC4–GFP was analyzed by confocal microscopy. The NPC4–GFP fusion protein localized to the rim of cells (Figure 7a), which is in agreement with plasma membrane localization as previously determined (Nakamura et al., 2005). Localization of the NPC5–GFP protein was different from that of NPC4–GFP. A diffuse intracellular fluorescence pattern very similar to that of a cytosolic GFP control was observed. Therefore, the NPC5 protein was tentatively localized to the cytosol. Given that the distinct localization patterns of the NPC4–GFP and NPC5–GFP fusion proteins were obtained by employing the same transformation vector and expression system and that the sequences are highly similar, it may be concluded that small amino acid differences must be responsible for protein allocation to the plasma membrane or cytosol, respectively.

Figure 7.

 Subcellular localization of NPC5.
(a) Fusion proteins of NPC4 and NPC5 with GFP under the control of the 35S promoter were transiently expressed in Arabidopsis WT leaves. Cytosolic control, pA7-GFP (GFP without signal sequence); ER control, KDEL-GFP. Localization was analyzed by confocal microscopy of GFP and chlorophyll fluorescence. Note that NPC4–GFP and NPC5–GFP localize to the plasma membrane and cytosol, respectively. Scale bars = 20 μm.
(b) The NPC5–GFP fusion protein (35S promoter) was stably expressed in npc5-1 mutant plants. Plants were grown on agar medium with (black bars) or without phosphate (grey bars), and lipids were measured by TLC and GC of fatty acid methyl esters (mean and SD, n = 3).
(c) Confocal microscopy of leaves of 35S:NPC5–GFP npc5-1 transgenic plants grown on soil reveals that green fluorescence is localized within the cells surrounding the chloroplast, indicative of cytosolic localization. Red fluorescence indicates chlorophyll. Scale bars = 20 μm.
(d) Arabidopsis WT plants were transformed with NPC4-GFP or NPCS-GFP constructs and raised on soil. Various subcellular fractions were isolated from leaves and used for Western blotting with anti-GFP antibodies. Lane 1, leaves; 2, chloroplasts; 3, microsomes; 4, plasma membrane; 5, ER/Golgi; 6, soluble protein.

The NPC4–GFP and NPC5–GFP fusion constructs were also stably expressed in Arabidopsis plants. Lipid measurements of transformed NPC5–GFP npc5-1 plants revealed that DGDG accumulated to approximately 27 mol% under phosphate deprivation, indicating that the npc5-1 mutation was complemented by introduction of the NPC5–GFP fusion protein (Figure 7b). Furthermore, analysis by confocal microscopy of npc5-1 leaf cells stably expressing NPC5–GFP confirmed that fluorescence was observed in a diffuse pattern within the cells surrounding the chloroplasts (cytosolic compartment) (Figure 7c).

Leaf extracts of transgenic plants expressing NPC4–GFP or NPC5–GFP were separated into various membrane fractions, and GFP fusion proteins were detected by Western blotting using anti-GFP antibodies (Figure 7d). The NPC4 fusion protein was highly abundant in the plasma membrane fraction, in accordance with the localization observed by confocal microscopy (Figure 7a,c) and previous results (Nakamura et al., 2005). In contrast, NPC5 was not detected in the plasma membrane fraction, but mostly localized to the soluble (cytosolic) fraction.


The replacement of phospholipids by glycolipids represents one of the most prominent changes in higher plants when subjected to phosphate limitation. During phosphate deprivation, the amounts of DGDG and SQDG increase at the expense of phospholipids, and expression of several genes involved in glycolipid synthesis is induced (Essigmann et al., 1998; Härtel et al., 2000; Kelly et al., 2003). We determined whether the transcription factor PHR1 is involved in regulation of membrane lipid changes. Expression of genes involved in galactolipid synthesis (MGD2, MGD3, DGD1, DGD2) and phospholipid degradation (PLDζ2, GPD1, NPC5) was upregulated in phosphate-deprived phr1 plants, although the upregulation was attenuated as compared to phosphate deprived WT (Figures 1 and 3). Direct analysis of lipid changes in phr1 demonstrated that PHR1 might contribute to a very minor extent to DGDG accumulation under low-phosphate conditions. Therefore, PHR1-independent pathways of regulation of glycolipid synthesis must exist. PHR1 belongs to a gene family of MYB transcription factors, and it is possible that other PHR1-like genes show overlapping functions and partially substitute for the loss of PHR1 activity in the phr1 mutant. An alternative mechanism for the regulation of membrane lipid changes under phosphate deprivation was shown to be based on the phytohormone auxin (Kobayashi et al., 2006).

The conversion of phospholipids to glycolipids requires the action of phospholipid degradation enzymes (Figure 2). Several genes encoding phospholipases or phospholipid-degrading enzymes are highly expressed during phosphate limitation. Expression of PLDζ2 is induced during phosphate limitation in roots (Cruz-Ramírez et al., 2006; Li et al., 2006a,b). The PLDζ subfamily of phospholipases D comprises two genes in Arabidopsis, PLDζ1 and PLDζ2 (Qin and Wang, 2002). No change in DGDG accumulation was detected in the leaves of pldζ1 or pldζ2 (Figure 4) (Cruz-Ramírez et al., 2006; Li et al., 2006b). Glycerophosphodiester phosphodiesterase (GPD) has previously been characterized at the biochemical level (van der Rest et al., 2002, 2004). GPD activity was localized to the extracellular space and to the vacuole, and was highly increased after phosphate deprivation. Expression of the GDP1 gene (At5g08030) is strongly upregulated during phosphate deprivation (Figure 3). Figure 4 shows that DGDG accumulation is not compromised in gpd1, indicating that GPD1 is not primarily involved in the replacement of phospholipids by glycolipids in leaves.

Arabidopsis contains a family of six non-specific phospholipase C genes. One gene, NPC4, was previously shown to be highly upregulated during phosphate deprivation and to encode the dominant phosphate-dependent phospholipase activity in leaves (Figure 6a) (Nakamura et al., 2005). In contrast to its high expression and high enzyme activity, no differences in DGDG accumulation in npc4 leaves were observed under phosphate deprivation (Figure 4) (Nakamura et al., 2005). The fact that NPC4 localizes to the plasma membrane suggests that it is involved in local phospholipid breakdown rather than bulk phospholipid degradation. Diacylglycerol produced by NPC4 might be employed for galactolipid production or could be involved in signaling. Arabidopsis contains a second, highly related gene (NPC5), which is adjacent to NPC4 on chromosome 3. This gene was previously not considered because its overall expression is very low. Figure 3 shows that NPC5 expression is induced in WT plants raised without phosphate and in pho1. More importantly, the increase in DGDG during phosphate deprivation was attenuated in npc5-1 leaves, and this result was confirmed by analyzing a second independent mutant allele, npc5-2, the pho1 npc5-1 double mutant, and npc5-1 mutant plants complemented with NPC5–GFP (Figures 4, 5b and 7b). As the extent of the changes in the phospholipid to glycolipid ratio are dependent on the experimental conditions, the pho1 npc5-1 double mutant system provides an alternative, reliable strategy to study the impact of phosphate deficiency on membrane lipid composition.

The DGDG accumulating under low-phosphate conditions in the npc5 mutant lines contains less 16:0 compared to WT (Table 1), suggesting that NPC5 is involved in the hydrolysis of extra-plastidial phospholipids, and that diacylglycerol released by the NPC5 reaction is used directly for galactolipid synthesis. Furthermore, NPC4 and NPC5 show a high degree of sequence similarity to phospholipases C of bacterial origin with specificity for the hydrolysis of bulk membrane phospholipids (Nakamura et al., 2005). Enzyme assays with recombinant proteins confirmed that NPC4 and NPC5 readily hydrolyze PC and PE, but NPC5 shows much lower activity (Figure 6c) (Nakamura et al., 2005). Taken together, these data suggest that NPC5 is involved in membrane phospholipid degradation rather than signaling.

NPC5 is responsible for approximately 50% of DGDG synthesis in leaves of Arabidopsis. The residual increase in DGDG observed in npc5 plants during phosphate deprivation might be due to the activity of alternative phospholipases involved in lipid turnover. Arabidopsis contains a large number of phospholipases organized into various gene families, some of which are induced during phosphate deprivation. Cruz-Ramírez et al. (2006) showed that DGDG production in roots under phosphate deprivation, but not that in leaves, is attenuated in the pldζ2 mutant. pldζ1, pldζ2 and the double mutant show no difference in DGDG accumulation under phosphate deprivation in leaves; a difference is seen only in the roots of pldζ1 pldζ2 (Li et al., 2006a,b). Similarly, we did not find any difference in DGDG accumulation in leaves of pldζ2 as compared to phosphate deprived WT (Figure 4). Of the two phospholipases Dζ, PLDζ2 shows the strongest induction of expression and the greatest impact on DGDG accumulation at low phosphate concentration (Li et al., 2006b; Cruz-Ramírez et al., 2006; Figure 2). To determine whether PLDζ2 is involved in the residual DGDG production in npc5-1 under phosphate limitation, an npc5-1 pldζ2 double mutant was generated. The DGDG content in npc5-1 pldζ2 grown under low-phosphate conditions was not further decreased compared to npc5-1 (Figure 4). Therefore, PLDζ2 is primarily involved in phosphate deprivation-induced lipid changes in roots (Cruz-Ramírez et al., 2006; Li et al., 2006b), while NPC5 is critical for lipid changes in leaves. Thus, alternative pathways might contribute to the change in the phospholipid to glycolipid ratio during phosphate deprivation, such as de novo synthesis of diacylglycerol at the ER or plastid from acyl ACP or acyl CoA and glycerol-3-phosphate, or the activity of alternative phospholipases or enzymes involved in phospholipid degradation or metabolism.

The cytosolic localization of the NPC5 enzyme might provide the means to target phospholipids in various cellular membranes simultaneously during phosphate limitation. Because phospholipids, the substrates of phospholipase C, are restricted to cellular membranes, the association of NPC5 with the membrane system is a prerequisite for phospholipid degradation (Figure 8). In agreement with this model, a minor fraction of NPC5 was always found to be associated with the endomembrane fraction (Figure 7d). Similarly to the induction of NPC5 gene expression during phosphate limitation, phospholipase D activity and expression are known to be upregulated during stress (e.g. wounding, elicitor treatment) or fruit ripening. A translocation from cytosolic localization to membrane association was shown for phospholipase D in tomato cells after elicitor treatment, in tomato fruits during maturation, and in castor leaves after wounding (Bargmann et al., 2006; Grittle Pinhero et al., 2003; Ryu and Wang, 1996). Therefore, translocation of phospholipases from the cytosol to the target membranes might represent a widespread mechanism for phospholipid degradation during signaling or lipid turnover. While the mechanism of enzyme binding to the lipid bilayer remains unclear, this model precludes the necessity for phospholipid transport to the sites of hydrolysis (Figure 8). Furthermore, this model implies that diacylglycerol is an important intermediate in phospholipid degradation during phosphate deprivation in leaves, and that diacylglycerol has to be transported to the plastid envelope membranes for subsequent conversion into MGDG and DGDG during phosphate deprivation.

Figure 8.

 Conversion of phospholipids to galactolipid in leaves by phospholipases NPC4 and NPC5 during phosphate deprivation.
The phospholipase NPC4 releases diacylglycerol from phospholipids during phosphate deprivation in the plasma membrane. NPC5 localizes to the cytosol and has access to various cellular membranes where it contributes to phospholipid degradation. Diacylglycerol in the outer chloroplast membrane can be used for MGDG and DGDG synthesis.

Experimental procedures

Arabidopsis mutant lines

The pho1-2 mutant (Delhaize and Randall, 1995), an allele of pho1-1 (Poirier et al., 1991), was obtained from the Nottingham Arabidopsis Seed Center (Nottingham, UK). The mutant genes in pho1-1 and pho1-2 carry premature stop codons, thus representing null alleles (Hamburger et al., 2002).

T-DNA insertional mutants for phr1 (SALK_067629; 11), pldζ2 (SALK_119084), gpd1 (SALK_133368) and npc5-1 (SALK_045037) were obtained from the Nottingham Arabidopsis Seed Center (ecotype Columbia). Isolation of the npc4 mutant has been described previously (Nakamura et al., 2005). A second mutant allele for npc5 (npc5-2, FLAG_483G02; ecotype Wassilewskija) was obtained from INRA Versailles (France). DNA sequences flanking the insertion sites were amplified by PCR. Sequencing of PCR products revealed that, for all mutants, the insertions localize to the genes between the start and stop codons. Homozygous lines were identified by PCR using gene-specific primers or combinations of gene-specific and T-DNA-specific primers (Table 2).

Table 2.   Oligonucleotides used in this study
PrimerDescriptionSequence (5′→3′)

Growth conditions

Arabidopsis seedlings were grown in liquid culture and deprived of phosphate for expression profiling experiments (Figure 2) as described by Bari et al. (2006) and Morcuende et al. (2007).

Plants were grown on 0.8% agar medium (Murashige and Skoog, 1962) containing 1% sucrose at a light intensity of 150 μmol m−2 sec−1 (16 h light per day) for 14 days, before being transferred to synthetic medium with or without phosphate (Figures 1b,c, 3b, 4, 6a and 7b and Table 1). The phosphate medium comprised 0.8% agarose, 1% sucrose, 2.5 mm KNO3, 1 mm MgSO4, 1 mm Ca(NO3)2, 1 mm NH4NO3, 1 mm KH2PO4, 25 μm Fe-EDTA, 35 μm H3BO3, 7 μm MnCl2, 0.25 μm CuSO4, 0.5 μm ZnSO4, 0.1 μm Na2MoO4, 5 μm NaCl and 5 nm CoCl2 (Estelle and Somerville, 1987); for phosphate-free medium, KH2PO4 was omitted. Leaf and root material for lipid or Northern analysis was harvested after 14 days of phosphate deprivation.

Wild-type, pho1 and npc mutant plants were raised on Murashige and Skoog (1962) medium containing 1% sucrose for the first 2 weeks after germination, and then grown on soil (Einheitserde GS90, Gebrüder Patzer, for an additional 3 weeks at 150 μmol m−2 sec−1 (16 h light per day). Leaves were harvested for lipid measurements, Northern blot or Western analysis (Figures 1b, 3b, 5 and 7c,d).

Northern blot analysis

Total RNA was isolated from frozen leaves, separated by agarose gel electrophoresis and blotted onto nylon membranes (Sambrook et al., 1989). The blots were hybridized with 32P-dCTP labeled probes derived from fragments obtained from genomic DNA by PCR. The 5′ UTRs were used as probes to distinguish expression levels of NPC4 and NPC5.

Lipid quantification and phospholipase C assays

Lipids were extracted from frozen leaves and lipids separated by TLC (Dörmann et al., 1995). Fatty acids from isolated lipids were converted into methyl esters, which were quantified by GC using pentadecanoic acid (15:0) as the internal standard (Browse et al., 1986).

The full-length NPC5 cDNA (At3g03540) was amplified by PCR using the primers NPC5-FpET and NPC5-RpET, and cloned into the pPICT2 vector (Kawaguchi et al., 2001; Nakamura et al., 2005). After sequencing, the cloned fragment was digested with EcoRV and ligated into the NdeI site of pET24a(+) (Novagen, Protein expression in E. coli was performed as described previously (Nakamura et al., 2005). NPC activities of recombinant protein (from E. coli) or of crude protein extracts from Arabidopsis leaves were measured using the radioactive substrates sn-1,2-di[1-14C]palmitoyl-PC and sn-1-palmitoyl-2-[1-14C]linoleoyl-PE, and product formation was recorded by TLC and autoradiography (Nakamura et al., 2005).

Subcellular localization of GFP fusion proteins by fluorescence microscopy

The cDNAs for NPC4 and NPC5 were ligated into the vectors pT7Blue and pPICT2, respectively, and, after SalI digestion, transferred into the SalI site of the GFP-35S vector, allowing expression of NPC4-GFP and NPC5-GFP under the control of the 35S promoter. GFP control constructs were obtained for the cytosol (pA7-GFP, GFP without signal sequence, provided by Bernd Müller-Röber, University of Potsdam, Germany) and ER (KDEL-GFP; Scott et al., 1999). Plasmid DNA was coated onto 1.0 μm gold micro-carriers (Bio-Rad, and transferred into leaf cells using a biolistic PSD-1000/He particle delivery system (Bio-Rad) with 7584kPa (1100 psi) rupture disks. The specimens were analyzed using a Leica DM-IRBE confocal fluorescence microscope (Leica,

The entire cassettes containing the NPC4–GFP or NPC5–GFP fusions were released from the GFP-35S vector and subcloned into the HindIII/EcoRI sites of the binary vector pBinAR-Hyg (Höfgen and Willmitzer, 1990). Binary constructs were transferred into Arabidopsis WT and npc5-1 via Agrobacterium-mediated transformation (Bent et al., 1994). Transformed plants expressing NPC4–GFP or NPC5–GFP were identified by Northern and Western blot analysis.

Isolation of subcellular compartments and Western blot

Subcellular compartments were isolated from leaves (20 g) of transgenic lines after homogenization in 300 ml extraction buffer (50 mm HEPES-KOH, pH 7.3, 330 mm sorbitol, 0.1% BSA) and filtration through Miracloth (Calbiochem, The homogenate was centrifuged (5 min at 2000 g). The pellet was re-suspended in extraction buffer and chloroplasts were isolated using a Percoll (GE Healthcare, step gradient (Ischebeck et al., 2006). Residual organelles were removed from the supernatant by centrifugation (10 000 g for 10 min). Microsomal membranes were separated from soluble proteins by centrifugation at 100 000 g for 60 min, and re-suspended in SPK buffer (330 mm sucrose, 5 mm K2HPO4, pH 7.8, 3 mm KCl). Membranes were separated by partitioning the microsomal suspension in a 5 g two-phase system containing 6.2% Dextran T 500 (Pharmacosmos, and 6.2% PEG 3350 (Sigma, in 330 mm sucrose, 5 mm K2HPO4, pH 7.8, 3 mm KCl, 2.5 mm NaCl. After mixing, phase separation was accomplished by centrifugation (1400 g for 5 min). The microsomal phase was subjected to three successive phase-partitioning steps. The upper phase, containing plasma membranes, and the lower phase, enriched in ER, Golgi and tonoplast, were diluted with SPK buffer. Membranes in the two diluted phases were each harvested by centrifugation at 100 000 g for 60 min.

Proteins were isolated from subcellular fractions using phenol (Cahoon et al., 1992), and used for polyacrylamide gel electrophoresis and Western blotting with nitrocellulose (Sambrook et al., 1989). Blots were probed with mouse monoclonal anti-GFP IgG1 (Invitrogen, and secondary antibodies (alkaline phosphatase-coupled goat anti-mouse antibody; Kirkegaard & Perry,, and signals were visualized using nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate p-toluidine salt (Roche Applied Science,


This work was partly supported by a Grant-in-Aid for Scientific Research on Priority Areas (number 18056007) from the Ministry of Education, Sports, Science and Culture in Japan. Financial support (Sonderforschungsbereich SFB429) from the Deutsche Forschungsgemeinschaft is gratefully acknowledged.