• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of plntcellLink to Publisher's site
Plant Cell. Mar 2005; 17(3): 760–775.
PMCID: PMC1069697

The Arabidopsis Plastidic Glucose 6-Phosphate/Phosphate Translocator GPT1 Is Essential for Pollen Maturation and Embryo Sac Development


Plastids of nongreen tissues can import carbon in the form of glucose 6-phosphate via the glucose 6-phosphate/phosphate translocator (GPT). The Arabidopsis thaliana genome contains two homologous GPT genes, AtGPT1 and AtGPT2. Both proteins show glucose 6-phosphate translocator activity after reconstitution in liposomes, and each of them can rescue the low-starch leaf phenotype of the pgi1 mutant (which lacks plastid phosphoglucoisomerase), indicating that the two proteins are also functional in planta. AtGPT1 transcripts are ubiquitously expressed during plant development, with highest expression in stamens, whereas AtGPT2 expression is restricted to a few tissues, including senescing leaves. Disruption of GPT2 has no obvious effect on growth and development under greenhouse conditions, whereas the mutations gpt1-1 and gpt1-2 are lethal. In both gpt1 lines, distorted segregation ratios, reduced efficiency of transmission in males and females, and inability to complete pollen and ovule development were observed, indicating profound defects in gametogenesis. Embryo sac development is arrested in the gpt1 mutants at a stage before the fusion of the polar nuclei. Mutant pollen development is associated with reduced formation of lipid bodies and small vesicles and the disappearance of dispersed vacuoles, which results in disintegration of the pollen structure. Taken together, our results indicate that GPT1-mediated import of glucose 6-phosphate into nongreen plastids is crucial for gametophyte development. We suggest that loss of GPT1 function results in disruption of the oxidative pentose phosphate cycle, which in turn affects fatty acid biosynthesis.


Nongreen plastids in heterotrophic tissues are carbohydrate-importing organelles, and in storage tissues amyloplasts are the site of starch synthesis. Nongreen plastids are normally unable to generate hexose phosphates from C3 compounds because they lack fructose 1,6-bisphosphatase activity (Entwistle and ap Rees, 1988), and they depend on the import of hexose phosphates as the source of carbon for starch biosynthesis and for the oxidative pentose phosphate pathway (OPPP). In all dicotyledonous species studied so far, nongreen plastids preferentially take up glucose 6-phosphate (Glc6P) as the hexose phosphate (Caspar et al., 1985; Kofler et al., 2000; for review, see Fischer and Weber, 2002). By contrast, in the endosperm of monocot cereals, starch biosynthesis preferentially employs ADP-glucose generated in the cytosol (Beckles et al., 2001), but these tissues also possess a Glc6P uptake system.

Indeed, the Glc6P/phosphate translocator (GPT) was first purified from plastid envelope membranes isolated from maize (Zea mays) endosperm (Kammerer et al., 1998), and corresponding cDNAs have since been isolated from various plant species (Kammerer et al., 1998). At the protein level, these GPTs display low, but significant, similarity to other subfamilies of plastidic phosphate antiport systems, namely, the triose phosphate/phosphate translocator (Flügge, 1999), the phosphoenolpyruvate/phosphate translocator (Fischer et al., 1997), and the xylulose 5-phosphate/phosphate translocator (Eicks et al., 2002). The proposed physiological function of the GPT is the Glc6P import into plastids of heterotrophic tissues for use as a precursor for starch (and fatty acid) biosynthesis and/or as a substrate for the OPPP. The OPPP provides reducing equivalents for biosynthetic pathways that rely on reducing power, such as the synthesis of fatty acids (Bowsher et al., 1992). These reducing equivalents are used for the reduction of 3-ketoacyl-ACP to acyl-ACP, a reaction catalyzed by two subunits of the fatty acid synthase complex, 3-ketoacyl reductase and enoyl-ACP reductase. The activity of fatty acid synthase ultimately leads to the synthesis of saturated 16:0 and 18:0 fatty acids that are used for the synthesis of various glycerolipids, such as phospholipids and triacylglycerol (for review, see Ohlrogge and Browse, 1995). Mutations that disrupt early steps in fatty acid biosynthesis are presumed to be lethal. Indeed, complete loss of function of the enzymes in this pathway is incompatible with viability (Mou et al., 2000; Carlsson et al., 2002; Baud et al., 2003).

In oilseed plants like Arabidopsis thaliana, a transient accumulation of starch during embryogenesis is followed by massive deposition of triacylglycerols, which are stored in oil bodies that account for ~60% of the cell volume in the cotyledons of mature embryos (Mansfield and Briarty, 1992). Changes in gene expression during seed development in Arabidopsis have been monitored by cDNA microarray analysis, and, interestingly, the expression of GPT1 mRNA was found to correlate with the transient accumulation of starch and to decline rapidly 8 d after flowering (Ruuska et al., 2002).

Just as storage oil bodies accumulate in cotyledons, pollen grains also accumulate intracellular lipid bodies, which act as reserves of energy and biomolecules for pollen germination. The pollen grain—the male or microgametophyte—develops within the anther in the stamen. The postmeiotic haploid microspore undergoes an asymmetric mitosis (pollen mitosis I), which results in a vegetative and a generative cell. The generative cell then undergoes a second mitosis to form two sperm cells (for review, see Mascarenhas, 1989; Twell et al., 1998; McCormick, 2004). Likewise, the female gametophyte, also referred to as the megagametophyte or embryo sac, develops within the ovule, which is found within the ovary in the carpel. Female gametophyte development comprises three successive rounds of nuclear division, followed by cellularization. The result is a seven-cell embryo sac that contains three antipodal, two synergid, one egg, and one central cell (reviewed in Reiser and Fischer, 1993; Drews et al., 1998; Yadegari and Drews, 2004). Mitochondrial and plastid DNAs are inherited maternally in Arabidopsis (Röbbelen, 1966; Martinez-Zapater et al., 1992; Martinez et al., 1997); the organellar DNAs in the male generative cell are degraded in a process that starts just after the first pollen mitosis (Nagata et al., 1999). In the pollen grain, therefore, intact plastids are found only in the vegetative cell. Because the components of the fatty acid biosynthetic machinery are located within plastids, intracellular lipid biosynthesis in pollen grains can only take place in the vegetative cell under the control of the gametophytic genome. Intracellular lipid biosynthesis gives rise to storage lipid bodies and an extensive membrane network (for review, see Piffanelli et al., 1998). Both the membrane and storage lipids of pollen grains provide the substrates for the rapid expansion of the plasmalemma after pollen germination and the subsequent elongation of the pollen tube.

The Arabidopsis genome contains two GPT genes, AtGPT1 (At5g54800) and AtGPT2 (At1g61800) that resulted from a duplication and a subsequent translocation event that were independent of the segmental duplications identified in Arabidopsis and its progenitors (Knappe et al., 2003a). In this study, we report on the functional characterization of the two AtGPT genes. Both GPTs are functional Glc6P translocators, but expression studies and the analysis of loss-of-function lines revealed that only GPT1 is essential. Mutations in gpt1 result in severe defects, especially during pollen development, whereas loss of GPT2 function has no obvious effect on plant development.


AtGPT1 and AtGPT2 Represent Functional Glc6P Translocators

To elucidate the functional characteristics of AtGPT1 and AtGPT2, cDNAs coding for the mature forms of both transporters were extended by a sequence coding for an N-terminal His6 affinity tag, cloned into the yeast expression vector pYES2 NT, and transformed into yeast cells. Affinity-purified AtGPT1 and AtGPT2 were then reconstituted into liposomes, which had been preloaded with different metabolites for the subsequent determination of transport specificities (Loddenkötter et al., 1993). The substrate specificities of the two transporters are listed in Table 1. Both Arabidopsis GPTs proved to be functional Glc6P transporters with almost identical substrate specificities, transporting inorganic phosphate, Glc6P, 3-phosphoglycerate, triose phosphates, and, to a lesser extent, phosphoenolpyruvate. In general, the transport specificities of the two Arabidopsis GPTs were similar to those of the GPT from pea (Pisum sativum) roots (PsGPT; Kammerer et al., 1998).

Table 1.
Substrate Specificities of Recombinant GPTs Expressed in Yeast Cells

The plastidic phosphoglucose isomerase (PGI; EC mutant pgi1 provided an opportunity to study AtGPT1 and AtGPT2 transport function in planta. The pgi1 mutant was isolated in a screen for leaf-specific starch deficiency (Yu et al., 2000). It retains only ~5% of wild-type plastid PGI activity (Yu et al., 2000), and its leaves contain ~25% of the wild-type level of starch (Figure 1). The plastid PGI converts Fru6P to Glc6P, which is then used as precursor for starch biosynthesis. Thus, if both GPTs represent functional Glc6P translocators, their ectopic expression in leaves should physiologically rescue the pgi1-specific starch deficiency phenotype by making cytosolic Glc6P available to chloroplasts as a precursor for starch biosynthesis. For this purpose, PsGPT, AtGPT1, and AtGPT2 cDNAs were stably expressed under the control of the 35S promoter in the pgi1-1 mutant background. For each construct, four to five independent lines were selected and analyzed for their leaf starch phenotype (Figure 1). The pgi1-1 35S:PsGPT lines showed a dramatic increase in leaf starch (130% of wild-type levels), indicating that PsGPT indeed delivers Glc6P to plastids for starch biosynthesis. The pgi1-1 35S:AtGPT1 and pgi1-1 35S:AtGPT2 lines displayed 80 and 135% of wild-type starch levels, respectively (Figure 1). Furthermore, the slight growth retardation characteristic of the pgi1-1 mutant was attenuated in each of the lines (data not shown). These experiments clearly demonstrate that both Arabidopsis proteins can deliver cytosolic Glc6P to the chloroplasts for starch biosynthesis, thus obviating the need for PGI. Therefore, both Arabidopsis GPTs represent functional GPTs in planta.

Figure 1.
Rescue of the Defect in Starch Accumulation in the pgi1-1 Mutant by Ectopic Expression of GPT.

Tissue-Specific Expression Patterns of AtGPTs

To investigate the expression of AtGPT1 and AtGPT2, publicly available microarray data were analyzed (https://www.genevestigator.ethz.ch/) (Zimmermann et al., 2004). AtGPT1 is ubiquitously expressed during development (e.g., in seeds, flowers, rosettes, and roots, with highest levels found in stamens) (Figure 2). In general, AtGPT2 is expressed at lower levels than AtGPT1; however, relatively higher levels were detected in sepals and senescing leaves (Figure 2). Similar expression patterns were observed using RT-PCR (data not shown).

Figure 2.
Digital Northern Analysis of AtGPT Genes.

To gain a deeper insight into the tissue-specific expression of AtGPT1, a translational fusion of the promoter sequence, the first exon and intron with the uidA gene, was stably introduced into Arabidopsis. AtGPT1:uidA expression was observed in leaf mesophyll and guard cells; β-glucuronidase (GUS) localization in mesophyll and guard cell chloroplasts was readily detectable (Figures 3A and 3B). To quantify GPT1 expression in these two leaf tissues, protoplasts of mesophyll and guard cells were prepared and subjected to RT-PCR. As shown in Figure 3C, a 10-fold increase in the accumulation of GPT1 transcripts was observed in guard cells relative to mesophyll cells. In roots, AtGPT1:uidA expression was restricted to the root cap, the vascular tissue, and the zone of lateral root formation (Figures 3D and 3E). In flowers, AtGPT1:uidA was expressed in pollen, as well as in the carpel, stigma, and the vascular tissue of sepals (Figure 3F). During early embryo development, AtGPT1:uidA expression was found throughout the whole seed, including the globular-stage embryo, the endosperm, and the integuments (Figures 3G and 3H). This ubiquitous expression pattern was also detected in early heart stage embryos (Figure 3I); however, during the late heart stage, AtGPT1:uidA expression began to disappear in the ground tissue (Figure 3J). At the transition to the torpedo stage, and in the mature embryo, AtGPT1:uidA expression was restricted to the root tip (Figures 3K and 3L). No expression was detected in mature seeds; only the pericarp showed weak AtGPT1:uidA expression (Figure 3M).

Figure 3.
GPT1:uidA Expression in Transgenic Arabidopsis Plants.

Isolation of Insertional Mutant Lines for AtGPT1

To identify A. thaliana lines that carry T-DNA insertions at the AtGPT1 locus, the Feldmann lines (Forsthoefel et al., 1992), the Arabidopsis Knockout Facility (AKF) BASTA population (Sussman et al., 2000), and the Salk Institute collection (Alonso et al., 2003) were screened by reverse genetics. In the Feldmann collection, two lines were identified, which harbored T-DNA insertions 586 bp upstream (gpt1-5) and 256 bp downstream (gpt1-1) of the start codon of AtGPT1, respectively (Figure 4A). In the BASTA population from the AKF, two lines were identified that contained T-DNA insertions at positions +1725 (gpt1-2) and +2754 (gpt1-6), relative to the ATG (Figure 4A). In addition, two lines (gpt1-3 and gpt1-4) were identified in the Salk collection, with T-DNA insertions at positions −19 and +1899 relative to the start codon of AtGPT1 (Figure 4A). To isolate homozygous mutant lines, different gene-specific primers were combined with the respective border primers for genotyping (Figure 4B). In the lines gpt1-3, gpt1-5, and gpt1-6, homozygous mutant plants were clearly identified (Figure 4B). However, RT-PCR analysis revealed that GPT1 expression was not impaired in these genotypes (data not shown). By contrast, no homozygous mutant lines could be obtained for gpt1-1, gpt1-2, or gpt1-4 (Figure 4B).

Figure 4.
Identification and Molecular Characterization of gpt1 Alleles.

Lines heterozygous for the gpt1-1 or the gpt1-2 mutation were chosen for further analysis. The heterozygous state of the lines was confirmed by DNA gel blot analysis using a gene-specific probe (Figure 4C). The filters were also hybridized with border-specific fragments. In the GPT1/gpt1-1 line, the right border–specific probe detected a 9-kb fragment only, indicating the presence of a single-copy T-DNA insertion. In GPT1/gpt1-2, the left border–specific probe detected a 1.8-kb fragment, but additional bands were also present, indicating the presence of several T-DNAs or a complex of insertions. This line was backcrossed to the wild type. In the subsequent F2 generation, it was not possible to isolate a single-copy line that showed BASTA resistance, so a PCR-based assay was used to monitor the segregation of the gpt1-2 allele in subsequent generations (see Methods).

Genetic Analysis and Complementation of gpt1

The progeny of line GPT1/gpt1-1, which contains a single-copy T-DNA insertion that confers kanamycin resistance (Kanr), segregated 0.9:1 for Kanr (Table 2). DNA gel blot analysis revealed that all Kanr plants contained the T-DNA insertion in GPT1 and were heterozygous at the GPT1 locus (data not shown). A total of 393 offspring of selfed GPT1/gpt1-2 plants was subjected to PCR analysis, and 176 lines were found to carry the gpt1-2 allele (Table 2), indicating a 0.8:1 segregation, very similar to the aberrant segregation observed for the gpt1-1 allele. The segregation ratios of both gpt1-1 and -2 were not significantly different from 1:1 (Table 2), instead of an expected 3:1, given the dominant mode of inheritance of both PCR and resistance phenotypes. The simplest explanation for the aberrant segregation patterns would be a gametophytic defect. To test this, male and female transmission efficiencies (TE) of the two defective gpt1 alleles were determined by performing reciprocal test crosses between heterozygous mutants and wild-type plants and analyzing the F1 progeny. Both gpt1-1 and gpt1-2 alleles showed reduced male and female TEs (Table 3); for gpt1-1, a female TE of 34% and a male TE of 20% was determined. Similar values were observed for the gpt1-2 allele: a female TE of 27% and a male TE of 20% in the test cross to Wassilewskija (Ws-2), and a female TE of 36% and a male TE of 24% in the test cross to Columbia (Col-0). Thus, the observed segregation ratio was not caused by a defect that is specific to the male or female line, but appeared to result from reduced TE through both gametophytes.

Table 2.
Segregation Analysis of GPT1/gpt1-1 and GPT1/gpt1-2
Table 3.
Analysis of Genetic Transmission of gpt1-1 and gpt1-2

Analysis of siliques of GPT1/gpt1-1 and GPT1/gpt1-2 plants revealed the presence of small, white, unfertilized ovules (Figure 5A) with a frequency of 32 and 28%, respectively, whereas such aborted ovules occurred with a frequency of only 7% in the wild type (Table 4). Although the less densely packed siliques of the heterozygous lines were shorter than those of the wild type (1.12 ± 0.12 cm versus 1.52 ± 0.08), the overall seed set (normal seeds plus aborted ovule) per silique did not differ markedly between heterozygous GPT1/gpt1-2 lines and wild-type plants (49 ± 5 versus 50 ± 7).

Figure 5.
Seed Phenotype of gpt1 Alleles and Genetic Complementation.
Table 4.
Frequency of Aborted Ovules and Pollen in the Wild Type, GPT1/gpt1-1, GPT1/gpt1-2, and in Two Complemented Lines

To complement the mutant phenotype, an 11-kb genomic fragment containing the GPT1 gene, including promoter and terminator regions, was introduced into GPT1/gpt1-2 plants. Nine independent T1 lines that inherited the gpt1-2 allele were generated. All of these lines were still heterozygous at the endogenous GPT1 locus (data not shown). Two lines (gpt1-2/GPT1 gGPT1-3 and gpt1-2/GPT1 gGPT1-9) were further analyzed in the T2 generation for segregation of the GPT1 locus using PCR. In each of the two T2 progenies, individual plants (8 out of 77 and 10 out of 28, respectively) were identified that were homozygous for the gpt1-2 allele. DNA from two representative plants, gpt1-2/gpt1-2 gGPT1-9.18 and gpt1-2/gpt1-2 gGPT1-3.10, was subjected to DNA gel blot analysis, and the homozygous state of the endogenous gpt1 locus was confirmed (Figure 5B). Siliques of gpt1-2/gpt1-2 gGPT1-9.18 and gpt1-2/gpt1-2 gGPT1-3.10 plants were inspected for aborted ovules in the T3 generation. In both lines, the frequency of aborted ovules was comparable to that in wild-type plants (Table 4, Figure 5C). This result confirms that the T-DNA insertions in the GPT1 gene were responsible for the lethal phenotype in the homozygous state and also increased the number of aborted ovules in siliques of heterozygous mutant plants.

Loss of GPT1 Function Perturbs Pollen Development

The reduction in TE via the male germ line in gpt1 plants suggests a role for GPT1 in pollen development and/or function. Pollen of wild-type and GPT1/gpt1 lines was analyzed for cytoplasmic density by the staining method of Alexander and for nuclear constitution by 4′,6-diamidino-2-phenylindole (DAPI) staining. In the Alexander assay, vital mature pollen grains show intensive red staining of the cytoplasm (Figure 6A). Dead and dying pollen, recognizable by its flattened shape and smaller size, exhibited only green staining of the pollen grain wall (Figure 6B). In wild-type plants, inviable pollen grains occurred with a frequency of <1%, whereas in GPT1/gpt1-1 and GPT1/gpt1-2 plants, the frequency was between 9 and 15% (Table 4). In both complemented lines, gpt1-2/gpt1-2 gGPT1-9.18 and gpt1-2/gpt1-2 gGPT1-3.10, the frequency of dead pollen was comparable to that in the wild type, confirming the importance of GPT1 for pollen development (Table 4). The small and flattened pollen grains were further investigated for their nuclear constitution by staining with DAPI. In wild-type pollen, the vegetative nucleus and the two generative nuclei were clearly distinguishable (Figure 6C). In the mutant pollen, only diffuse staining was observed (Figure 6D). Besides this strong mutant phenotype, we observed several intermediate phenotypes in DAPI- and Alexander-stained pollen, which might result in inviability after release (data not shown).

Figure 6.
Loss of GPT1 Function Perturbs Pollen Development.

Using scanning electron microscopy, pollen from wild-type and heterozygous gpt1 lines was examined more closely. Pollen grains from the wild type appeared normal (Figure 6E), whereas significant numbers of abnormal and collapsed pollen grains were detected in the GPT1/gpt1 lines (Figure 6F), which is consistent with our light microscopic observations (Figures 6A to 6D). Although the mutant pollen grains were smaller in size and less robust, the ordered reticulate pattern of the wild-type exine layer was still prominent (Figures 6G and 6H), indicating that the basic structure of the exine layer is not altered. The exine layer is produced by the sporophytic tissue, which suggests that the heterozygous tissue of the sporophyte has no influence on the development of gpt1 mutant pollen.

Pollen Grains Harboring the gpt1 Allele Have Fewer Vacuoles and Lipid Bodies and Show a Tendency to Lyse

To define the time point of gpt1 pollen grain collapse, the expression of GPT1:uidA was investigated. GPT1:uidA expression was readily detected in tricellular pollen grains (Figures 7A and 7B), whereas GUS activity could not be unequivocally demonstrated in uninuclear und bicellular pollen (data not shown). This GPT1:uidA expression profile is in good agreement with publicly available microarray data (Figure 7C), which show that GPT1 expression levels are highest in tricellular pollen.

Figure 7.
GPT1 Expression in Pollen Grains.

Cross sections of wild-type and GPT1/gpt1-2 anthers, analyzed by transmission electron microscopy, revealed that development of the pollen sac, including its tapetum layer, was similar in the two genotypes (data not shown). The gpt1-2 defect was evident in pollen sacs of GPT1/gpt1-2 anthers (anther stage 12, according to Sanders et al., 1999) harboring mature tricellular pollen grains (Figure 8A). At this stage, the tapetal cell layer has almost completely disappeared, and degeneration of the septum gives rise to a bilocular anther. Within the locule of the pollen sac, a mixture of pollen grains was found. Wild-type–like pollen grains within the GPT/gpt1-2 locule were identified by the presence of lipid bodies and vacuoles (Figure 8B). The mutant pollen grains showed varying degrees of damage and could be grouped into three classes (weak, moderate, and strong). Weakly affected mutant pollen still has intracellular structures but lacks vacuoles (Figure 8C). In addition, significantly fewer lipid bodies were present in cross sections of mutant pollen (25 ± 5.8) compared with 72 ± 15.4 lipid bodies in cross sections of wild-type pollen (t test, P < 0.001). In agreement with the observations made by light microscopy, a significant proportion of mutant pollen grains were collapsed (Figure 8D). The endothecial cells of the heterozygous GPT1/gpt1-2 tissue contained plastids with starch granules and plastoglobuli and therefore exhibited a wild-type–like phenotype (Figure 8D). Similar observations were made with the GPT1/gpt1-1 line (data not shown).

Figure 8.
Pollen Sacs from GPT1/gpt1-2 Plants Contain a Mixture of Normal and Mutant Pollen Grains.

Besides the prominent constituents of wild-type pollen grains, namely, the lipid bodies and the vacuoles (possibly storage vacuoles, as suggested in Yamamoto et al., 2003), smaller vesicles were detected throughout the cytoplasm (Figure 9A). In addition, plastids containing starch granules were frequently observed in wild-type pollen grains (Figures 9B and 9C). Figures 9D to 9F show that weakly affected mutant pollen still contained the vegetative nucleus, mitochondria, and even occasional lipid bodies. However, the plastid-like structures appeared to be free of starch granules (Figure 9F). Even in these weakly affected pollen grains, autolysis was obvious at several sites. In moderately affected pollen, autolysis was more extensive, and the organelles in the vegetative cell seemed to be clustered (Figures 9G and 9H), although distinguishable mitochondria and Golgi stacks were still present. Vacuole-like structures were observed only very rarely (Figure 9I). Large inclusions were detected in all mutant pollen grains (Figures 9D, 9G, and 9J). In strongly affected pollen, autolysis leads to the disintegration of basic structures (Figures 9J and 9K). This process was accompanied by a swelling of the intine wall and retraction of the plasma membrane (Figure 9L). Despite the severity of the damage in the vegetative cells, the exine layer of these pollen grains was still essentially wild-type (Figures 9C and 9L), as confirmed by scanning electron microscopy (Figure 6H).

Figure 9.
Ultrastructural Alterations in gpt1 Mutant Pollen.

The gpt1 Lesion Arrests Embryo Sac Development

As outlined above, siliques of both the GPT1/gpt1-2 and GPT1/gpt1-1 lines contained ~30% aborted ovules (Table 4), and in both lines, the TE of the gpt1 allele was reduced. The actual frequency of aborted ovules is in good agreement with the value expected (average of 32%) on the basis of the transmission data (Table 3). Detection of GPT1:uidA expression within the embryo sac (Figure 10A) is compatible with a role for GPT1 during female gametophyte development. We therefore investigated the nuclear constitution of the embryo sacs in affected ovules 48 h after the emasculation of GPT1/gpt1-2 flowers. In ovaries of wild-type plants, all embryo sacs reached the terminal developmental stage (containing one secondary nucleus, one egg cell nucleus, and two synergid cell nuclei). However, in ovaries of GPT1/gpt1-2 plants, only 64% (n = 117) of embryo sacs displayed the wild-type phenotype (Figure 10B). The majority of mutant embryo sacs were consistently found to possess two slightly larger nuclei lying side by side and to display a variable number of smaller nuclei (Figure 10C). This phenotype resembles that of the gfa2 mutant described by Christensen et al. (2002), in which the polar nuclei fail to fuse. In that study, the polar nuclei migrated properly, came to lie side by side, and were slightly enlarged. Thus, it is most likely that gpt1 mutant embryo sacs are defective in nuclear fusion. A small fraction of the remaining embryo sacs was found to be in an advanced state of degeneration, containing only one nucleus (Figure 10D). After pollination, the two larger polar nuclei persisted in mutant embryo sacs until the stage at which the proembryo and the suspensor became distinguishable in wild-type seeds (data not shown).

Figure 10.
GPT1:uidA Expression in Ovules and the Terminal Developmental Stage of gpt1-2 Embryo Sacs.

GPT2 Function Is Not Essential for Gametogenesis

Although these data indicate that the GPT1 protein is essential for efficient gametophyte development, a fraction of the gametophytes still remains functional, indicating incomplete penetrance of the mutation. Therefore, the question arises whether GPT2 can partly substitute for GPT1. The expression data for GPT2 revealed low expression levels in stamens and carpels (Figure 2). However, no significant GPT2 expression was found in uninuclear, bicellular, or tricellular pollen grains (data from the Nottingham Arabidopsis Stock Centre [NASC] at the Genevestigator Web site, https://www.genevestigator.ethz.ch/). Furthermore, a line was identified in the GABI-Kat collection (Li et al., 2003), which contained a T-DNA insertion at position +1433 relative to the start codon (gpt2-1). No GPT2 transcripts are detectable in homozygous gpt2-1 mutants by RT-PCR (data not shown). These lines exhibited essentially wild-type growth and development under greenhouse conditions (data not shown). Because a slight reduction in fertility might not have been detectable under these conditions, the mutant flowers were examined by microscopy for pollen and ovule aberrations. Neither the frequency of aborted pollen (0.4%, n = 6400), as revealed by the Alexander stain, nor that of aborted ovules (4%, n = 716) differed from that observed in the wild type. Together, the data suggest that the GPT2 function has only minor influence on pollen and embryo sac development.


In Planta Function of AtGPT1 and AtGPT2

The GPTs represent one of the four known subfamilies of plastidic phosphate translocators. The Arabidopsis genome contains two functional GPT genes, AtGPT1 and AtGPT2 (Knappe et al., 2003a), whose products share 75% sequence identity. Both translocators show similar substrate specificities, accepting Glc6P and triose phosphates as countersubstrates for inorganic phosphate, and these specificities are similar to those of pea GPT (Table 1). The mode of GPT action is a 1:1 exchange of Glc6P mainly with inorganic phosphate and triose phosphates or an exchange of triose phosphates with inorganic phosphate (Kammerer et al., 1998). In addition, when ectopically expressed, both GPTs can rescue the low-starch leaf phenotype of the pgi1 mutant, indicating that both proteins are able to deliver Glc6P to chloroplasts as a precursor for starch biosynthesis, thereby circumventing the necessity to convert Fru6P to Glc6P (the reaction catalyzed by PGI). Thus, both GPTs function as glucose-6-phosphate transporters in planta. However, their expression patterns differ markedly. Whereas GPT1 is ubiquitously expressed, GPT2 is expressed at a lower level in most tissues. Remarkably, GPT2 is highly expressed in response to Pseudomonas syringae treatment (data not shown) and in a few tissues, such as senescing leaves (Figure 2). Whether AtGPT2 has a specific function during pathogen treatment and senescence remains to be determined. On the other hand, the expression pattern of AtGPT1 implies that it is the major GPT responsible for the transport of Glc6P into plastids of heterotrophic tissues in Arabidopsis. It is expressed in roots, as is the pea GPT (Borchert et al., 1993; Kammerer et al., 1998), and AtGPT1 expression is strongest in the root cap, the site of starch synthesis. In the pgi1 mutant, starch synthesis in roots is not affected (Yu et al., 2000), most probably because Glc6P can be imported into amyloplasts via AtGPT1, thus bypassing the need for PGI. In leaves, AtGPT1 is expressed in guard cells and, to a lesser extent, also in mesophyll cells (Figures 3A to 3C). The expression in guard cells is in accordance with previous findings that guard cell chloroplasts lack fructose-1,6-bisphosphatase activity and therefore rely on the availability of hexose phosphate for starch biosynthesis (Hedrich et al., 1985). The subsequent breakdown of starch yields malate as a counter ion for potassium during stomatal opening. In contrast with mesophyll chloroplasts, envelope membranes of guard cell chloroplasts have been shown to have GPT transport activity (Overlach et al., 1993). The AtGPT1 expression observed in Arabidopsis leaf mesophyll cells is somewhat surprising because the chloroplasts of these cells do not depend on the import of reduced carbon sources and Glc6P transport activities are not detectable (data not shown). Also, the low-starch leaf phenotype of the pgi1 mutant obviously cannot be compensated for by the endogenous AtGPT1 activity. It is conceivable that expression of GPT1 mRNA in these cells serves as a standby system (i.e., that synthesis or activation of the GPT1 protein operates only on demand). This assumption is corroborated by the observation that, upon feeding with glucose, a specific Glc6P transport system is induced in spinach (Spinacia oleracea) leaves leading to starch accumulation (Quick et al., 1995).

AtGPT1 Is Crucial for Normal Male Gametogenesis

No obvious abnormalities could be detected in heterozygous gpt1 lines; indeed, the mutation was expected to be recessive. However, the phenotypic and genetic analyses revealed a major role of GPT1 in the haploid phase of Arabidopsis development, namely, during embryo sac and pollen development.

Given the apparent physiological role of GPT1—import of Glc6P into plastids—the question arises why a defect in Glc6P transport should cause such a severe phenotype as that observed in gpt1 pollen. Glc6P that enters the plastid via the GPT1 can be used (1) as a carbon source for starch biosynthesis, (2) as a substrate for fatty acid biosynthesis, or (3) as a starter molecule for the OPPP. The role of GPT1 in delivering the precursor for starch biosynthesis is obvious because plastids harboring starch granules were frequently found in wild-type pollen grains (Figures 9B and 9C). By contrast, the mutant gpt1 pollen only contained starch-free plastids (Figure 9F). However, the existence of several starch-free mutants of Arabidopsis (Lin et al., 1988; Kofler et al., 2000) that retain full fertility and display no reduction in pollen viability (P. Niewiadomski and A. Schneider, unpublished results) demonstrates that starch accumulation per se is not a prerequisite for pollen development.

Glc6P or even triose phosphates can also be used as a carbon skeleton for fatty acid biosynthesis. However, fatty acid biosynthesis also can be driven by a range of alternative substrates (e.g., phosphoenolpyruvate, acetate, malate, or pyruvate) (for review, see Fischer and Weber, 2002). Therefore, it remains unclear to what extent Glc6P and triose phosphates might serve as precursors for fatty acid biosynthesis. Nevertheless, we assume that GPT1 fulfils an essential role in fatty acid metabolism during pollen development. Fatty acid biosynthesis in the plastids depends on the availability of reducing equivalents, which are generated by the first enzymatic reactions in the OPPP. An inadequate supply of Glc6P as a precursor for the OPPP, therefore, would restrict activity of the fatty acid synthase complex. It has been shown that a mutation in enoyl-ACP reductase, which catalyzes the final reaction in the de novo fatty acid biosynthesis cycle and consumes reducing equivalents, leads to premature cell death and altered morphology in Arabidopsis (Mou et al., 2000). Although the mod1 mutation, a single amino acid substitution in the enoyl-ACP reductase, allowed residual enzyme activity and caused a reduction of only ~10% in lipid content, the pleiotropic morphological effects in this mutant were severe (Mou et al., 2000). A role for GPT1 in fatty acid metabolism during pollen development is supported by the following observations: GPT1 expression increases during pollen development (Figure 7), as one would expect if the GPT1 function has to meet increasing demands for reducing equivalents for fatty acid biosynthesis. In developing pollen grains, oil body accumulation, as well as the proliferation of an extensive network of endoplasmic reticulum (ER) membranes and vesicles, have been reported to commence after the first pollen mitosis (Piffanelli et al., 1998) Likewise, the pattern of lipid marker gene expression coincides with the accumulation of intracellular storage and membrane lipid components in Brassica napus pollen (Piffanelli et al., 1997). Furthermore, while weakly affected gpt1 pollen grains still possessed lipid bodies, the overall number of lipid bodies in mutant pollen was significantly reduced. The variation in lipid body content and in pollen viability might be caused by variable rates of GPT1 protein depletion, with functional GPT1 protein presumably being carried over from plastids of the microspore mother cell. There was no indication for the involvement of GPT2, although the possibility that GPT2 is activated in the absence of GPT1 cannot be excluded.

A defect in lipid biosynthesis should also affect the integrity of cell membranes. The membrane-bound glycerol-3-phosphate acyltransferase (GPAT) mediates the initial step in glycerolipid biosynthesis in extraplastidic compartments of plant cells. The Arabidopsis gpat1 mutant displays defects in ER membrane biogenesis and altered oil body size in pollen grains; gpat1 pollen grains, moreover, show reduced competitiveness in fertilization (Zheng et al., 2003). Surprisingly, in weakly and moderately affected gpt1 pollen, the ER network and other membrane systems, like the nuclear membrane, the Golgi apparatus, and mitochondria, were not visibly altered compared with the wild type. Only in strongly affected gpt1 pollen grains was a disorganization of the plasma membrane observed. This phenomenon, however, may be considered as a secondary effect because it was associated with a swelling of the pollen wall and occurred only at an advanced state of autolysis.

Another striking feature of gpt1 mutant pollen was the almost complete absence of dispersed vacuoles in tricellular pollen (Figures 8C and 9D to 9L). In the wild type, the vacuoles undergo serial transformations during pollen development (e.g., in microspores, a large vacuole appears, which divides into small vegetative vacuoles after the first mitosis) (Yamamoto et al., 2003). These vegetative vacuoles disappear after the second mitosis, and in mature pollen grains, storage vacuoles appear that are produced de novo from the rough ER. Finally, in pollen grains from flowers in bloom, these storage vacuoles are converted into lytic vacuoles (Yamamoto et al., 2003). In all microspores of heterozygous gpt1 plants, the large vacuole was intact (data not shown). In tricellular gpt1 pollen grains, however, vacuoles were nearly completely absent (Figures 8C and 9D to 9I). The absence of vacuoles in gpt1 pollen might be explained as a direct consequence of reduced fatty acid biosynthesis, which will lead to a relative lack of components for the formation of the vacuole membranes. Interestingly, a mutation in the VCL1 gene, which is essential for vacuole biogenesis, affects pollen performance after germination, yet it is not lethal in the gametophyte (Hicks et al., 2004).

In summary, we propose the following working model for the function of GPT1 in pollen development: Developing pollen has a high demand for fatty acids. In gpt1 pollen, lack of Glc6P results in a reduced supply of reducing equivalents via the OPPP, which directly affects the formation of lipid bodies and vacuoles. The decrease in the level of lipid bodies and of vacuoles is not lethal per se, but the vegetative cell may sense the severe impairment of fatty acid biosynthesis and might initiate an autolytic process leading to nonphysiological cell death. This model is in accordance with the finding that interruption of any vital metabolic process inevitably leads to nonphysiological cell death (Vaux and Korsmeyer, 1999). The model is also supported by the observation that the onset of autolysis precedes the disintegration of mitochondria, rough ER, Golgi stacks, and nuclear membranes (Figures 9D and 9G).

Reducing equivalents are generated via the OPPP in reactions catalyzed by glucose-6-phosphate 1-dehydrogenase (G6PDH; EC and 6-phoshogluconate dehydrogenase (6PGDH; EC In the case that the OPPP plays a pivotal role during pollen development, genes encoding these enzymes should be expressed in these tissues. The expression pattern of genes, which are likely to encode plastidic isoforms of G6PDH and 6PGDH (Kruger and von Schaewen, 2003) are summarized in Table 5. Of the four genes coding for G6PDH and the two genes coding for 6PGDH, the expression pattern of the 6PGDH gene At3g02360 closely resembled that of GPT1 (i.e., highest expression was found in tricellular pollen grains). In addition, an insertion mutant of this gene (At3g02360) produced a high number (32%) of aborted pollen and could not be established in the homozygous state, as is the case for gpt1 (P. Niewiadomski and A. Schneider, unpublished results). This indicates that the product of At3g02360 might be crucial for both the OPPP and pollen development, suggesting a functional relationship between the oxidative part of the OPPP and pollen development.

Table 5.
Expression Analysis of Genes Involved in Plastidic OPPP Metabolism

Involvement of AtGPT1 in Embryo Sac and Seed Development

In addition to its essential role during pollen maturation, GPT1 is important for normal embryo sac development. Thus, the gpt1 mutation affected the fusion of the polar nuclei during embryo sac development (Figure 10C). In the wild type, fusion of the polar nuclei involves fusion of outer and inner nuclear membranes, and the fusion of the outer nuclear membrane has been suggested to be dependent on the mitochondrial GFA2 protein (Christensen et al., 2002). Although it is premature to speculate on the underlying mechanism, AtGPT1 clearly also has a gametophytic effect on female fertility. An involvement of the OPPP seems possible because plastids of the female gametophyte might also depend on the import of reduced carbon sources via GPT1 for the generation of reducing equivalents.

Furthermore, although 5 to 9% homozygous gpt1 lines are predicted based on TEs measured, we never recovered a homozygous adult plant. Therefore, future work will involve the generation of lines in which the gametophytic defects of the gpt1 mutation are removed (e.g., by complementing the gpt1 mutant with GPT1 driven by a pollen-specific and/or embryo sac–specific promoter). This would allow us to investigate the effects of loss of GPT1 function in the sporophyte, in particular during seed development. It is worth noting that GPT:uidA-based (see Figures 3G to 3L) and microarray-based analyses have shown that the expression of GPT1 in seeds covaries with the one of starch metabolizing enzymes (Ruuska et al., 2002). Therefore, it will be interesting to see if GPT1 plays a role during the starch accumulating phase and/or during the oil accumulating phase throughout embryogenesis.


Plant Material

Seeds of Arabidopsis thaliana (Heynh.) (ecotypes Ws-2 [N1601] and Col-0 [N1093], a collection of 6500 T-DNA–transformed lines obtained by seed transformation [Feldmann collection; Forsthoefel et al., 1992] and arranged in pool sizes of 100 [N3115 and N6500] and 20 [N3116 and N6400], and two lines identified at the Salk Institute [N521762 and N589293]) were provided by the NASC (http://nasc.nott.ac.uk/home.html). Seeds of pools 2, D, plate 66 and 8, F, plate 67 were obtained from the AKF (http://www.biotech.wisc.edu/Arabidopsis/), and seeds of line 454A06 were supplied by GABI-Kat (http://www.mpiz-koeln.mpg.de/GABI-Kat/). Plants were grown in a temperature-controlled greenhouse in a light/dark cycle of 16 h/8 h or in a growth chamber in a light/dark cycle of 12 h/12 h at day/night temperatures of 21°C/18°C and 40% humidity.

Construction of Promoter:uidA and 35S:cDNA Fusions and Analysis of Transgenic Plants

The GPT1-promoter-GUS construct was generated as a translational fusion by ligating a DNA fragment comprising the promoter and the 5′ end of the coding region of GPT1 to the uidA gene. The AtGPT1 DNA fragment (2.1 kb) was generated using a proofreading DNA polymerase (Pfx; Invitrogen, Karslruhe, Germany) and the primer combination GPT1PromF/GPT1PromR (for primer sequences, see Table 6) and subcloned into pBluescript KS− (pBS; Stratagene, La Jolla, CA). After restriction with SalI/fill in-XbaI, the DNA fragment was cloned into SmaI+XbaI-digested pGPTV-bar vector (Becker et al., 1992), resulting in pGPT1prom. The 35S:cDNA constructs were assembled starting from the appropriate cDNAs. The AtGPT1 cDNA was identified by screening an Arabidopsis cDNA library (Clontech, Palo Alto, CA) and subsequently subcloned into pBS (Stratagene) to give pGPT1. The AtGPT2 cDNA was amplified by RT-PCR using the gene-specific primers GPT2-F1/GPT2-R1 (Table 6) and subcloned into pGEM-Teasy (Promega, Madison, WI), resulting in pGPT2. The 35S:GPT constructs were generated by cloning the SacII/fill in-SalI fragments of pGPT1 and pGPT2 into the SmaI+SalI-digested pBinAR-Kan vector (Höfgen and Willmitzer, 1990), resulting in p35SAtGPT1 and p35SAtGPT2, respectively. The 35S:PsGPT construct was generated by cloning the SmaI-SalI fragment of PsGPT-cDNA (Kammerer et al., 1998) into the SmaI+SalI-digested pBinAR-Kan, resulting in p35SPsGPT. Transgenic plants were generated by vacuum infiltration of Arabidopsis plants using Agrobacterium tumefaciens cultures containing the appropriate construct (Bechtold et al., 1993). The primary transformants were allowed to flower and produce seeds.

Table 6.
Primers Used in This Study

Transformants were selected with BASTA or kanamycin and verified by PCR analysis. Histochemical localization of GUS in transgenic plants harboring the GPT1:uidA construct was performed as described previously (Knappe et al., 2003b). The starch content of leaves from control plants and transgenics harboring 35S:cDNA constructs was determined by the method described by Lin et al. (1988).

Isolation of gpt1 and gpt2 Alleles

The Feldmann collection was subjected to PCR using primers F-LB, F-RB, and GPT1-R5 (for primer sequences, see Table 6) as described previously (Schneider et al., 2002). The AKF BASTA population was screened for gpt1 insertion alleles using a primer specific for the T-DNA left border (JL-202) in combination with GPT1-Fw (see http://www.biotech.wisc.edu/arabidopsis for Methods). Appropriate PCR bands were identified by DNA gel blot analysis using the AtGPT1 cDNA as a probe, subcloned, and sequenced. In silico analysis of the SALK and GABI-Kat databases (http://signal.salk.edu/cgi-bin/tdnaexpress and http://www.mpiz-koeln.mpg.de/GABI-Kat/) led to the identification of a gpt2 and additional gpt1 insertion lines.

Genotyping of lines was done using a PCR-based approach. Genomic DNA was isolated (Liu et al., 1995) from each plant analyzed and used as a template for PCR amplification of DNA fragments corresponding to the wild-type alleles and the insertion alleles. For heterozygous lines, the following primer combinations were used: for the gpt1-1 line, F-RB/GPT1-R5 and GPT1-F2/GPT1-R5; for the gpt1-2 line, Jl-202/GPT1-F3 and GPT1-F3/GPT1-R3, and for the gpt1-4 line, S-LB/GPT1-R2 and GPT1-R2/GPT1-F3 (see Figure 4B; for primer sequences, see Table 6). For the homozygous gpt1-3 line, S-LB/GPT1-F5 and GPT1-F2/GPT1-R4 were used, for gpt1-5, F-LB/GPT1-R4 and GPT1-F1/GPT1-R6, and for gpt1-6, Jl-202/GPT1-F4 and GPT1-F4/GPT1-R1 (see Figure 4B, Table 6). The gpt2-1 line (GABI-Kat line 454A06) was genotyped using the GPT2-specific primer combination GPT2-F2/GPT2-R2 and the T-DNA–specific primer GK-LB together with GPT2-R2. DNA gel blots (10 μg of DNA) were prepared and analyzed following standard protocols (Sambrook et al., 1989) using a 645-bp GPT1, a 708-bp RB, or a 299-bp LB fragment as the probe (Figure 4C).

Genetic Analysis

To examine gametophytic transmission of the gpt1-1 (background Ws-2) and gpt1-2 (background Ws-2) alleles, reciprocal test crosses were performed between wild-type (Ws-2 and Col-0) and mutant lines. Seeds harvested from crosses of gpt1-1 with Ws-2 were sown on kanamycin-containing plates, and the resistance phenotype was scored. Seeds obtained from crosses of gpt1-2 with Ws-2 or Col-0 were sown on soil, and genomic DNAs from the F1 progeny were analyzed by PCR using the primer combination Jl-202/GPT1-F3. The success of crosses between gpt1-2 and Col-0 was monitored using the cleaved-amplified polymorphic sequence marker ER (http://www.arabidopsis.org/; Konieczny and Ausubel, 1993). No cases of adventitious self-fertilization were detected among the F1 progeny. The TE of T-DNA via each type of gamete (TE male and TE female) was calculated as described previously (Howden et al., 1998). The percentage of arrested gametophytes based on transmission data was calculated according to the following formula: (100 − background)/2 − TE × (100 − background)/2 = % aborting events.

Phenotypic Analysis and Quantification

Seed development was analyzed in the 10th to 12th siliques of the main inflorescence of self-pollinated heterozygous gpt1 plants and wild-type plants at a stage when the final silique size had been reached. To determine the terminal phenotype of mutant ovules, flowers were emasculated and the ovule phenotype was analyzed 48 h after emasculation. Flowers were fixed for 16 h in an ethanol:acetic acid mixture (9:1), washed in 80 and 70% ethanol, and cleared in chloral hydrate:H2O:glycerol (8:3:1).

Analysis of mature pollen with DAPI was performed as previously described (Park et al., 1998) Alexander staining of pollen was performed on released pollen grains by adding them directly to Alexander's stain (Alexander, 1969).

Preparations were examined with a light microscope (Eclipse E800; Nikon, Tokyo, Japan) equipped for differential interference contrast and fluorescence microscopy. Images were captured using a 1-CCD color video camera (KY-F1030; JVC, Singapore) operated by the DISKUS software package (Technisches Büro Hilgers, Königswinter, Germany).

Transmission and Scanning Electron Microscopy

Stamens for transmisssion electron microscopy were taken from different flower stages and fixed with 2% glutaraldehyde in 50 mM phosphate buffer, pH 7.4, overnight at 4°C. Samples were postfixed in 1% osmium tetroxide for 8 h on ice, dehydrated in a graduated acetone series, including a step with 1% uranylacetate (in 50% acetone, 2 h), embedded in Spurr's resin, and polymerized at 50°C for ~72 h. Ultrathin sections (60 to 70 nm) were cut with a diamond knife (Micro Star, Huntsville, TX) on a Leica Ultracut UCT microtome (Leica Microsystems, Vienna, Austria) and mounted on pioloform-coated copper grids. The sections were stained with lead citrate and uranyl acetate (Reynolds, 1963) and viewed with a Zeiss EM 109 transmission electron microscope (Carl Zeiss, Oberkochen, Germany) at 80 kV. Micrographs were taken using SO-163 EM film (Kodak, Rochester, NY).

For scanning electron microscopy, released pollen grains were mounted on stubs and sputter-coated with gold particles (S150A; Edwards, Crawley, UK). Specimens were examined with a scanning electron microscope (XL 30 ESEM; Philips, Eindhoven, The Netherlands) at an accelerating voltage of 15 kV.

Genetic Complementation

A genomic library of Arabidopsis in λEMBL3 (Clontech) was screened by plaque hybridization using the AtGPT1 cDNA insert as a probe following standard protocols (Sambrook et al., 1989). A genomic DNA clone containing the GPT1 locus was isolated and an 11-kb SalI-BamHI fragment was subcloned into pBS (Stratagene) and sequenced. The SalI-BamHI fragment, including the GPT1 coding region flanked by 3.3 kb of 5′ and 5.3 kb of 3′ noncoding sequence, was excised from the cloning vector and inserted into the binary vector pGreen (Hellens et al., 2000), resulting in pGreen-GPT1. pGreen-GPT1 was introduced into A. tumefaciens strain 3101 containing pSoup (Hellens et al., 2000) and subsequently transformed into GPT1/gpt1-2 plants (see Figure 5). Transgenic Arabidopsis plants were selected for kanamycin resistance. Segregation analysis of the endogenous GPT1 locus was performed by PCR using the primer combination GPT1-F6/GPT1-R7 (Table 6) and TaKaRa Ex-Taq polymerase according to the manufacturer's instructions (BioWhittaker, Verviers, Belgium). Generation of a 5-kb fragment indicates the presence of the endogenous GPT1 locus and not the introduced genomic GPT1 fragment.

Expression Analysis

Total RNA was isolated from different mutant lines as previously described (Eggermont et al., 1996). Oligo(dT)-primed cDNA from 2 μg of total RNA (DNase treated) was synthesized using the SuperScript reverse transcriptase system (Invitrogen). Primers used for amplification were GPT1RT-F/GPT1RT-R and GPT2RT-F/GPT2RT-R (for primer sequences, see Table 6), and reactions were performed for 5 min at 95°C followed by 40 cycles of 30 s at 94°C, 1 min at 55°C, and 2 min at 72°C.

For the isolation of guard cell and mesophyll cell protoplasts, epidermal peels and epidermis-free developed rosette leaves were incubated for 2 h or 20 to 30 min, respectively, in solutions containing 0.8% (w/v) cellulase (Onozuka R-10; Serva, Heidelberg, Germany), 0.1% pectolyase (Sigma-Aldrich, St. Louis, MO), 0.5% BSA, 0.5% polyvinylpyrrolidone, 1 mM CaCl2, and 10 mM Mes/Tris, pH 5.6. The osmolarity of the enzyme solution was adjusted to 540 (for guard cell protoplasts) or 400 (for mesophyll cell protoplasts) mosmol kg−1 with d-sorbitol. Protoplasts released from guard cells or mesophyll tissues were recovered by filtration through a 20-μm nylon mesh and washed twice in 1 mM CaCl2 buffer. Protoplast RNA was purified twice with the Dynabeads mRNA Direct kit (Dynal, Oslo, Norway) to minimize DNA contamination. Quantitative real-time RT-PCR was performed in a Light Cycler (Roche, Mannheim, Germany) as described before (Ivashikina et al., 2003) using AtACT2/8fwd/AtACT2/8rev and GPT1RT-F/GPT1RT-R as primer pairs (for primer sequences, see Table 6).

In silico expression analysis was performed at https://www.genevestigator.ethz.ch/.

Heterologous Expression of AtGPTs in Yeast Cells and Reconstitution of Transport Activities

For heterologous expression in yeast cells, cDNAs coding for the mature forms of AtGPT1 and AtGPT2 were cloned into the yeast expression vector pYES2-NT (Invitrogen). The mature AtGPT1 was obtained by cloning the PvuI/blunted-EcoRI fragment of pGPT1 into BamHI/fill-in+EcoRI-digested pYES2-NTC, resulting in pYESmGPT1. The mature form of AtGPT2 was obtained by cloning the PvuI/blunted-EcoRI fragment of pGPT2 into BamHI/fill-in+EcoRI-digested pYES2-NTA, resulting in pYESmGPT2. Both constructs were transformed into the Saccharomyces cerevisiae strain INVSc1 (Invitrogen), and transformants were selected using URA3 as the selective marker. The induction of protein expression was done according to the manufacturer's instructions. Yeast cells were harvested 6 h after induction and disrupted in 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 300 μg mL−1 phenylmethylsulfonylfluoride. The 100,000g yeast membrane fractions were prepared by ultracentrifugation and solubilized using 2% (w/v) n-dodecyl maltoside, and the recombinant His6-GPT1 and His6-GPT2 proteins were then purified via metal affinity chromatography on Ni2+-nitrilotriacetic acid agarose (Qiagen, Hilden, Germany) and used for reconstitution of transport activities. Reconstitution of whole tissue extracts and of transport activities was performed as previously described (Loddenkötter et al., 1993; Kammerer et al., 1998; Knappe et al., 2003b).


We thank the Genomic Analysis Laboratory of the Salk Institute for providing the sequence-indexed Arabidopsis T-DNA insertion mutants, Kerstin Kunze, Hildegard Voll, Rita Grotjahn, and Barbara Hess for technical assistance, Siegfried Werth for photographs, Rita Gross-Hardt (University of Zürich, Switzerland) for support with ovule analysis, and Dario Leister (Max-Planck-Institute for Plant Breeding, Cologne, Germany), Paul Hardy (University of Düsseldorf, Germany), and Eric van der Graaff for critical reading of the manuscript. This work was supported by grants from the Deutsche Forschungsgemeinschaft, the Bundesministerium für Bildung und Forschung, and the Fonds der Chemischen Industrie.


The author responsible for distribution of materials integral to the findings presented in this article in accordance with policy described in the Instructions for Authors (www.plantcell.org) is: Anja Schneider (anja.schneider/at/uni-koeln.de).

Article, publication date, and citation information can be found at www.plantcell.org/cgi/doi/10.1105/tpc.104.029124.


  • Alexander, M.P. (1969). Differential staining of aborted and nonaborted pollen. Stain Technol. 44, 117–122. [PubMed]
  • Alonso, J.M., et al. (2003). Genome-wide insertional mutagenesis of Arabidopsis thaliana. Science 301, 653–657. [PubMed]
  • Baud, S., Guyon, V., Kronenberger, J., Wuilleme, S., Miquel, M., Caboche, M., Lepiniec, L., and Rochat, C. (2003). Multifunctional acetyl-CoA carboxylase 1 is essential for very long chain fatty acid elongation and embryo development in Arabidopsis. Plant J. 33, 75–86. [PubMed]
  • Bechtold, N., Ellis, J., and Pelletier, G. (1993). In planta Agrobacterium mediated gene transfer by infiltration of adult Arabidopsis thaliana plants. C. R. Acad. Sci. 316, 1194–1199.
  • Becker, D., Kemper, E., Schell, J., and Masterson, R. (1992). New plant binary vectors with selectable markers located proximal to the left T-DNA border. Plant Mol. Biol. 20, 1195–1197. [PubMed]
  • Beckles, D.M., Smith, A.M., and ap Rees, T.A. (2001). A cytosolic ADP-glucose pyrophosphorylase is a feature of graminaceous endosperms, but not of other starch-storing organs. Plant Physiol. 125, 818–827. [PMC free article] [PubMed]
  • Borchert, S., Harborth, J., Schünemann, D., Hoferichter, P., and Heldt, H.W. (1993). Studies of the enzymic capacities and transport properties of pea root plastids. Plant Physiol. 101, 303–312. [PMC free article] [PubMed]
  • Bowsher, C.G., Boulton, E.L., Rose, J., Nayagam, S., and Emes, M.J. (1992). Reductant for glutamate synthase is generated by the oxidative pentose-phosphate pathway in nonphotosynthetic root plastids. Plant J. 2, 893–898.
  • Carlsson, A.S., LaBrie, S.T., Kinney, A.J., von Wettstein-Knowles, P., and Browse, J. (2002). A KAS2 cDNA complements the phenotypes of the Arabidopsis fab1 mutant that differs in a single residue bordering the substrate binding pocket. Plant J. 29, 761–770. [PubMed]
  • Caspar, T., Huber, S.C., and Somerville, C. (1985). Alterations in growth, photosynthesis, and respiration in a starchless mutant of Arabidopsis thaliana (L.) deficient in chloroplast phosphoglucomutase activity. Plant Physiol. 79, 11–17. [PMC free article] [PubMed]
  • Christensen, C.A., Gorsich, S.W., Brown, R.H., Jones, L.G., Brown, J., Shaw, J.M., and Drews, G.N. (2002). Mitochondrial GFA2 is required for synergid cell death in Arabidopsis. Plant Cell 14, 2215–2232. [PMC free article] [PubMed]
  • Drews, G.N., Lee, D., and Christensen, C.A. (1998). Genetic analysis of female gametophyte development and function. Plant Cell 10, 5–17. [PMC free article] [PubMed]
  • Eggermont, K., Goderis, I.J., and Broekaert, W.F. (1996). High-throughput RNA extraction from plant samples based on homogenisation by reciprocal shaking in the presence of a mixture of sand and glass beads. Plant Mol. Biol. Rep. 14, 273–279.
  • Eicks, M., Maurino, V., Knappe, S., Flügge, U.I., and Fischer, K. (2002). The plastidic pentose phosphate translocator represents a link between the cytosolic and the plastidic pentose phosphate pathways in plants. Plant Physiol. 128, 512–522. [PMC free article] [PubMed]
  • Entwistle, G., and ap Rees, T.A. (1988). Enzymic capacities of amyloplasts from wheat (Triticum aestivum) endosperm. Biochem. J. 255, 391–396. [PMC free article] [PubMed]
  • Fischer, K., Kammerer, B., Gutensohn, M., Arbinger, B., Weber, A., Häusler, R.E., and Flügge, U.I. (1997). A new class of plastidic phosphate translocators: A putative link between primary and secondary metabolism by the phosphoenolpyruvate/phosphate antiporter. Plant Cell 9, 453–462. [PMC free article] [PubMed]
  • Fischer, K., and Weber, A. (2002). Transport of carbon in non-green plastids. Trends Plant Sci. 7, 345–351. [PubMed]
  • Flügge, U.I. (1999). Phosphate translocators in plastids. Annu. Rev. Plant Physiol. Plant Mol. Biol. 50, 27–45. [PubMed]
  • Forsthoefel, N.R., Wu, Y., Schulz, B., Bennett, M.J., and Feldmann, K.A. (1992). T-DNA insertion mutagenesis in Arabidopsis: Prospects and perspectives. Aust. J. Plant Physiol. 19, 353–366.
  • Hedrich, R., Raschke, K., and Stitt, M. (1985). A role for fructose-2,6-bisphosphate in regulating carbohydrate metabolism in guard cells. Plant Physiol. 79, 977–982. [PMC free article] [PubMed]
  • Hellens, R.P., Edwards, E.A., Leyland, N.R., Bean, S., and Mullineaux, P.M. (2000). pGreen: A versatile and flexible binary Ti vector for Agrobacterium-mediated plant transformation. Plant Mol. Biol. 42, 819–832. [PubMed]
  • Hicks, G.R., Rojo, E., Hong, S., Carter, D.G., and Raikhel, N.V. (2004). Geminating pollen has tubular vacuoles, displays highly dynamic vacuole biogenesis, and requires VACUOLESS1 for proper function. Plant Physiol. 134, 1227–1239. [PMC free article] [PubMed]
  • Höfgen, R., and Willmitzer, L. (1990). Biochemical and genetic analysis of different patatin isoforms expressed in various organs of potato (Solanum tuberosum). Plant Sci. 66, 221–230.
  • Howden, R., Park, S.K., Moore, J.M., Orme, J., Grossniklaus, U., and Twell, D. (1998). Selection of T-DNA-tagged male and female gametophytic mutants by segregation distortion in Arabidopsis. Genetics 149, 621–631. [PMC free article] [PubMed]
  • Ivashikina, N., Deeken, R., Ache, P., Kranz, E., Pommerrenig, B., Sauer, N., and Hedrich, R. (2003). Isolation of AtSUC2 promoter-GFP-marked companion cells for patch-clamp studies and expression profiling. Plant J. 36, 931–945. [PubMed]
  • Kammerer, B., Fischer, K., Hilpert, B., Schubert, S., Gutensohn, M., Weber, A., and Flügge, U.I. (1998). Molecular characterization of a carbon transporter in plastids from heterotrophic tissues: The glucose 6-phosphate/phosphate antiporter. Plant Cell 10, 105–117. [PMC free article] [PubMed]
  • Knappe, S., Flügge, U.I., and Fischer, K. (2003. a). Analysis of the plastidic phosphate translocator gene family in Arabidopsis and identification of new phosphate translocator-homologous transporters, classified by their putative substrate-binding site. Plant Physiol. 131, 1178–1190. [PMC free article] [PubMed]
  • Knappe, S., Löttgert, T., Schneider, A., Voll, L., Flügge, U.I., and Fischer, K. (2003. b). Characterization of two functional phosphoenolpyruvate/phosphate translocator (PPT) genes in Arabidopsis: AtPPT1 may be involved in the provision of signals for correct mesophyll development. Plant J. 36, 411–420. [PubMed]
  • Kofler, H., Häusler, R.E., Schulz, B., Gröner, F., Flügge, U.I., and Weber, A. (2000). Molecular characterisation of a new mutant allele of the plastid phosphoglucomutase in Arabidopsis, and complementation of the mutant with the wild-type cDNA. Mol. Gen. Genet. 263, 978–986. [PubMed]
  • Konieczny, A., and Ausubel, F.M. (1993). A procedure for mapping Arabidopsis mutations using co-dominant ecotype-specific PCR-based markers. Plant J. 4, 403–410. [PubMed]
  • Kruger, N.J., and von Schaewen, A. (2003). The oxidative pentose phosphate pathway: Structure and organisation. Curr. Opin. Plant Biol. 6, 236–246. [PubMed]
  • Li, Y., Rosso, M.G., Strizhov, N., Viehoever, P., and Weisshaar, B. (2003). GABI-Kat SimpleSearch: A flanking sequence tag (FST) database for the identification of T-DNA insertion mutants in Arabidopsis thaliana. Bioinformatics 19, 1441–1442. [PubMed]
  • Lin, T.-P., Caspar, T., Somerville, C., and Preiss, J. (1988). Isolation and characterization of a starchless mutant of Arabidopsis thaliana (L.) lacking ADPglucose pyrophosphorylase activity. Plant Physiol. 86, 1131–1135. [PMC free article] [PubMed]
  • Liu, Y.G., Mitsukawa, N., Oosumi, T., and Whittier, R.F. (1995). Efficient isolation and mapping of Arabidopsis thaliana T-DNA insert junctions by thermal asymmetric interlaced PCR. Plant J. 8, 457–463. [PubMed]
  • Loddenkötter, B., Kammerer, B., Fischer, K., and Flügge, U.I. (1993). Expression of the functional mature chloroplast triose phosphate translocator in yeast internal membranes and purification of the histidine-tagged protein by a single metal-affinity chromatography step. Proc. Natl. Acad. Sci. USA 90, 2155–2159. [PMC free article] [PubMed]
  • Mansfield, S.G., and Briarty, L.G. (1992). Cotyledon cell-development in Arabidopsis thaliana during reserve deposition. Can. J. Bot. 70, 151–164.
  • Martinez, P., Lopez, C., Roldan, M., Sabater, B., and Martin, M. (1997). Plastid DNA of five ecotypes of Arabidopsis thaliana: Sequence of ndhG gene and maternal inheritance. Plant Sci. 123, 113–122.
  • Martinez-Zapater, J.M., Gil, P., Capel, J., and Somerville, C.R. (1992). Mutations at the Arabidopsis CHM locus promote rearrangements of the mitochondrial genome. Plant Cell 4, 889–899. [PMC free article] [PubMed]
  • Mascarenhas, J.P. (1989). The male gametophyte of flowering plants. Plant Cell 1, 657–664. [PMC free article] [PubMed]
  • McCormick, S. (2004). Control of male gametophyte development. Plant Cell 16 (suppl.), S142–S153. [PMC free article] [PubMed]
  • Mou, Z., He, Y., Dai, Y., Liu, X., and Li, J. (2000). Deficiency in fatty acid synthase leads to premature cell death and dramatic alterations in plant morphology. Plant Cell 12, 405–418. [PMC free article] [PubMed]
  • Nagata, N., Saito, C., Sakai, A., Kuroiwa, H., and Kuroiwa, T. (1999). The selective increase or decrease of organellar DNA in generative cells just after pollen mitosis one controls cytoplasmic inheritance. Planta 209, 53–65. [PubMed]
  • Ohlrogge, J., and Browse, J. (1995). Lipid biosynthesis. Plant Cell 7, 957–970. [PMC free article] [PubMed]
  • Overlach, S., Diekmann, W., and Raschke, K. (1993). Phosphate translocator of isolated guard-cell chloroplasts from Pisum sativum L. transports glucose-6-phosphate. Plant Physiol. 101, 1201–1207. [PMC free article] [PubMed]
  • Park, S.K., Howden, R., and Twell, D. (1998). The Arabidopsis thaliana gametophytic mutation gemini pollen1 disrupts microspore polarity, division asymmetry and pollen cell fate. Development 125, 3789–3799. [PubMed]
  • Piffanelli, P., Ross, J.H., and Murphy, D.J. (1997). Intra- and extracellular lipid composition and associated gene expression patterns during pollen development in Brassica napus. Plant J. 11, 549–562. [PubMed]
  • Piffanelli, P., Ross, J.H., and Murphy, D.J. (1998). Biogenesis and function of the lipidic structures of pollen grains. Sex. Plant Reprod. 11, 65–80.
  • Quick, W.P., Scheibe, R., and Neuhaus, H.E. (1995). Induction of hexose-phosphate translocator activity in spinach chloroplasts. Plant Physiol. 109, 113–121. [PMC free article] [PubMed]
  • Reiser, L., and Fischer, R.L. (1993). The ovule and the embryo sac. Plant Cell 5, 1291–1301. [PMC free article] [PubMed]
  • Reynolds, E.S. (1963). The use of lead citrate at high pH as an electron-opaque stain in electron microscopy. J. Cell Biol. 17, 208–212. [PMC free article] [PubMed]
  • Röbbelen, G. (1966). Chloroplast differentiation following gene induced plastom mutation in Arabidopsis thaliana (L.). Heynh. Z. Pflanzenphysiol. 55, 387–403.
  • Ruuska, S.A., Girke, T., Benning, C., and Ohlrogge, J.B. (2002). Contrapuntal networks of gene expression during Arabidopsis seed filling. Plant Cell 14, 1191–1206. [PMC free article] [PubMed]
  • Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual, 2nd ed. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press).
  • Sanders, P.M., Bui, A.Q., Weterings, K., McIntire, K.N., Hsu, Y., Lee, P.Y., Truong, M.T., Beals, T.B., and Goldberg, R.B. (1999). Anther developmental defects in Arabidopsis thaliana male-sterile mutants. Sex. Plant Reprod. 11, 297–322.
  • Schneider, A., Häusler, R.E., Kolukisaoglu, U., Kunze, R., van der Graaff, E., Schwacke, R., Catoni, E., Desimone, M., and Flügge, U.I. (2002). An Arabidopsis thaliana knock-out mutant of the chloroplast triose phosphate/phosphate translocator is severely compromised only when starch synthesis, but not starch mobilisation is abolished. Plant J. 32, 685–699. [PubMed]
  • Sussman, M.R., Amasino, R.M., Young, J.C., Krysan, P.J., and Austin-Phillips, S. (2000). The Arabidopsis knockout facility at the University of Wisconsin-Madison. Plant Physiol. 124, 1465–1467. [PMC free article] [PubMed]
  • Twell, D., Park, S.K., and LaLanne, E. (1998). Asymmetric division and cell-fate determination in developing pollen. Trends Plant Sci. 3, 305–310.
  • Vaux, D.L., and Korsmeyer, S.J. (1999). Cell death in development. Cell 96, 245–254. [PubMed]
  • Yadegari, R., and Drews, G.N. (2004). Female gametophyte development. Plant Cell 16 (suppl.), S133–S141. [PMC free article] [PubMed]
  • Yamamoto, Y., Nishimura, M., Hara-Nishimura, I., and Noguchi, T. (2003). Behavior of vacuoles during microspore and pollen development in Arabidopsis thaliana. Plant Cell Physiol. 44, 1192–1201. [PubMed]
  • Yu, T.S., Lue, W.L., Wang, S.M., and Chen, J. (2000). Mutation of Arabidopsis plastid phosphoglucose isomerase affects leaf starch synthesis and floral initiation. Plant Physiol. 123, 319–326. [PMC free article] [PubMed]
  • Zheng, Z., Xia, Q., Dauk, M., Shen, W., Selvaraj, G., and Zou, J. (2003). Arabidopsis AtGPAT1, a member of the membrane-bound glycerol-3-phosphate acyltransferase gene family, is essential for tapetum differentiation and male fertility. Plant Cell 15, 1872–1887. [PMC free article] [PubMed]
  • Zimmermann, P., Hirsch-Hoffmann, M., Hennig, L., and Gruissem, W. (2004). GENEVESTIGATOR. Arabidopsis microarray database and analysis toolbox. Plant Physiol. 136, 2621–2632. [PMC free article] [PubMed]

Articles from The Plant Cell are provided here courtesy of American Society of Plant Biologists


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Gene
    Gene links
  • GEO Profiles
    GEO Profiles
    Related GEO records
  • HomoloGene
    HomoloGene links
  • MedGen
    Related information in MedGen
  • Pathways + GO
    Pathways + GO
    Pathways, annotations and biological systems (BioSystems) that cite the current article.
  • Protein
    Published protein sequences
  • PubMed
    PubMed citations for these articles
  • Taxonomy
    Related taxonomy entry
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...