Virus-induced gene silencing of plastidial soluble inorganic pyrophosphatase impairs essential leaf anabolic pathways and reduces drought stress tolerance in Nicotiana benthamiana.

The role of pyrophosphate in primary metabolism is poorly understood. Here, we report on the transient down-regulation of plastid-targeted soluble inorganic pyrophosphatase in Nicotiana benthamiana source leaves. Physiological and metabolic perturbations were particularly evident in chloroplastic central metabolism, which is reliant on fast and efficient pyrophosphate dissipation. Plants lacking plastidial soluble inorganic pyrophosphatase (psPPase) were characterized by increased pyrophosphate levels, decreased starch content, and alterations in chlorophyll and carotenoid biosynthesis, while constituents like amino acids (except for histidine, serine, and tryptophan) and soluble sugars and organic acids (except for malate and citrate) remained invariable from the control. Furthermore, translation of Rubisco was significantly affected, as observed for the amounts of the respective subunits as well as total soluble protein content. These changes were concurrent with the fact that plants with reduced psPPase were unable to assimilate carbon to the same extent as the controls. Furthermore, plants with lowered psPPase exposed to mild drought stress showed a moderate wilting phenotype and reduced vitality, which could be correlated to reduced abscisic acid levels limiting stomatal closure. Taken together, the results suggest that plastidial pyrophosphate dissipation through psPPase is indispensable for vital plant processes.


INTRODUCTION
Pyrophosphate (PP i ) is a key metabolite generated in the activation of several polymerization steps (Geigenberger et al., 1998;Stitt, 1998;Rojas-Beltrán et al., 1999;Farré et al., 2001;Sonnewald, 2001;López-Marqués et al., 2004), and its removal is essential to prevent the inhibition of thermodynamically unfavourable reactions (Geigenberger et al., 1998;López-Marqués et al., 2004). PP i is generally removed by inorganic pyrophosphatases which hydrolyze PP i to orthophosphate (P i ). Pyrophosphatases are ubiquitous in plant cells and found both as soluble forms in the cytosol and plastid, as well as membrane-bound forms on the tonoplast (Rea and Poole, 1993;Baltscheffsky et al., 1999;Maeshima, 2000), mitochondria (Vianello and Macrí, 1999) and chloroplast (Jiang et al., 1997). In Arabidopsis thaliana six soluble pyrophosphatase (sPPase) isoforms have been identified to date (Schulze et al., 2004). Five (AtPPa1,2,3,4 and 5) are far more similar to each other than to AtPPa6 (Schulze et al., 2004), and have been shown to be localized to the cytosol using GFP fusions (Ergen, 2006). In potato, two sPPase genes, StPPa1 and -2 which are similar to AtPPa1, have also been identified and demonstrated to be present in the cytosol using immunogold labeling (Rojas-Beltrán et al., 1999). In addition to sPPases, several other cytosolic enzymes can remove PP i , including the soluble enzymes pyrophosphate:fructose 6-phosphate phosphotransferase (PFP) and UDP-glucose pyrophosphorylase (UGPase). Due to a lack of adverse phenotypic alterations found when altering the expression of these enzymes in autotrophic sink metabolism, a considerable activity resides in the plastid (Gross and ap Rees, 1986;Weiner et al., 1987;Gómez-Garcia et al., 2006) and relatively low PP i levels are therefore maintained (5-15% of the total cellular PP i content is in the plastid compared with approximately 70% in the cytosol (Weiner et al., 1987;Farré et al., 2006)), the most likely explanation is that basal plastidial sPPase activity tightly governs and efficiently hydrolyzes plastidial PP i . In Arabidopsis a single isoform, AtPPa6, with an N-terminal plastid transit peptide extension has been identified (Schulze et al., 2004). Subsequent import (Schulze et al., 2004) and GFP localization studies (Ergen, 2006) have confirmed its subcellular localization. Plastidial PP i generation occurs during several metabolic pathways within the chloroplast, for example chlorophyll, starch, nucleic acid, carotenoid/xanthophyll (see Fig. S1), fatty acid and amino acid biosynthesis, and it is hypothesized that these pathways could be severely inhibited if PP i is not effectively removed.
Recent studies on key processes in primary metabolism have demonstrated that much still has to be learned about its regulation (Crevillén et al., 2003;Kulma et al., 2004;Kolbe et al., 2005;Sparla et al., 2005;Lunn et al., 2006;Marri et al., 2009;Petreikov et al., 2010) and understanding the influence of pyrophosphate would expand our knowledge further. Here we demonstrate that a transient repression of the native plastidial sPPase gene using virus induced gene silencing (VIGS) led to increased PP i, levels associated with altered starch, chlorophyll, carotenoid, malate and histidine content, as well as affected photosynthesis in N.

VIGS repression of plastidial sPPase activity in N. benthamiana leaves
Sequence analysis of the DNA sequence of At5g09650 (AtPPa6) revealed high similarity (75.5%) of the Solanum lycopersicum EST clone cLET20N17 (Genebank acc. AW092511.1) to the AtPPa6 cDNA sequence (Schulze et al., 2004), sharing only between 25.4% and 28.1% identity to the cytosolic targeted AtPPa1-5 sequences (Fig. S2), thus strongly suggesting that the tomato EST encodes a plastidial sPPase (psPPase). This cDNA was used to produce a vector that is able to induce VIGS of the sPPase encoded by it through ligation of the fragment into the multiple cloning site of the deconstructed TRV2 vector (Liu et al., 2002a). Subsequent co-infiltration with TRV1 and TRV2 (empty or containing the cLET20N17 fragment) into N. benthamiana seedlings resulted in reduced total in gel sPPase activity (Fig. 1A) and an approximate 90% reduction in total sPPase maximal catalytic activity when compared to the TRV2 empty vector controls (Fig.   1C). In order to examine whether the repression was specific to the plastidial isoform, fractions enriched in chloroplast marker enzymes (Table S1) were also subjected to an in-gel assay for sPPase activity. A band corresponding to those of the plastidial fraction was significantly reduced in total protein extracts of psPPase silenced plants (Fig. 1A, Fig. S3A). Similarly, an immunoblot using antibodies specifically recognizing the cytosolic isoforms (Rojas-Beltrán et al. 1999) was performed ( Fig. 1B) and showed a band of approximately 30 kDa that was similar in intensity of extracts from both TRV2 control and psPPase silenced tobacco plants, demonstrating that the VIGS-inhibition was specific to psPPase. Lastly, total pyrophosphate content from leaves of psPPase silenced and TRV2 control plants was measured and indicated a threefold increase in cellular PP i in the silenced plants compared to the controls (Fig. 1D). A striking observation in the TRV2-psPPase plants was the appearance of mottling on the source leaves compared with the control ( Fig. 2A, B). This suggests that pigment accumulation was significantly affected. To further investigate this, chlorophyll, carotenoid and xanthophyll contents were analyzed by high pressure liquid chromatography (HPLC) ( Table I). This indicated that the β -carotene, chlorophyll a (chl a) and violaxanthin contents of psPPase silenced plants were reduced between 30 and 50% compared to the controls (Table I). In contrast, chlorophyll b (chl b) and lutein remained invariable from the control, whilst zeaxanthin content was increased three fold in the silenced plants (Table I).

Effect of reduced psPPase activity on photosynthesis and
Carbon assimilation was severely impaired in the psPPase silenced plants ( Fig. 2C). At both 380 and 1000 µmol.mol -1 CO 2 concentrations carbon fixation was significantly reduced by 62 and 71% respectively in the down regulated psPPase leaves (Fig. 2C). Starch amounts were also measured and showed a significant decrease in the silenced leaves (Table II). In order to distinguish between a decrease in starch content resulting from increased PP i levels or decreased photosynthesis rates, down regulated psPPase plants were dark-adapted for three days to allow complete degradation of starch in the leaves. Subsequently, the leaf discs were transferred onto sucrose, kept in the dark, and starch content measured 6 hrs and 24 hrs after supplementation. Starch levels increased significantly in the TRV2 control plants; while down regulated psPPase plants could not synthesize starch under these conditions (Fig. 3A). In-gel activity assays for phosphoglucomutase (PGM) showed no discernable difference between silenced and control plants (Fig. S3C), while ADP glucose pyrophosphorylase (AGPase) activity was increased by approximately 40% in the silenced plants (Fig. 3B).
Lastly, soluble sugars levels (sucrose, glucose and fructose) did not change in comparison with the TRV2 control (Table II).

Effect of reduced sPPase activity on protein expression and metabolite level
In light of the fact that a reduction in sPPase led to a significant reduction in plastidial carbon metabolism, it was decided to also investigate protein and metabolite content following the accumulation of pyrophosphate. Total soluble protein content expressed on equal leaf area basis was significantly reduced by approximately 60% in the psPPase plants compared to the controls (Fig. 3C).
Furthermore denaturing protein gels were evaluated for differences in protein banding pattern between TRV2 control and psPPase silenced leaves. Interestingly, the major discernable differences were in the accumulation of both the nuclear and plastidial encoded Rubisco subunits which were greatly reduced in the down regulated psPPase plants (Fig. 3D). To determine whether the observed effect was due to inhibition of transcription semi-quantitative RT-PCR expression analysis of the two subunits was performed (Fig. 3E). Relative mRNA accumulation values indicated a 30% reduction in the plastidially encoded rbcL subunits with no significant change in the nuclear encoded rbcS subunits (Fig. 3E).
Non-redundant primary metabolites identified via gas chromatography time of flight mass spectrometry (GC-TOF-MS) indicated that glyceraldehyde 3phosphate, malate, quinate, myo-inositol, tryptophan, histidine, serine, ferulate, coniferylalcohol, 3-caffeoyl quinate, 4-caffeoyl quinate, 5-caffeoyl quinate and 1pyrroline-2-carboxylate levels significantly increased in psPPase silenced plants compared to the respective control under well-watered conditions. In contrast, citrate levels decreased significantly under the same conditions (Fig. 4, for full list see Table S2). In light of the fact that an alteration in PP i levels could lead to several secondary effects, a linear correlation matrix between prevailing PP i levels and the complete subset of primary metabolites were constructed. This indicated highly significant positive correlations between PP i levels and ferulate, coniferylalcohol, malate, quinate, histidine, glyceraldehyde 3-phosphate and 5caffeoyl quinate content (r>0.8), while significant negative correlations between PP i levels and chl a and 1 0 synthesis and starch accumulation. However, nucleotide levels measured remained either unaltered or were only slightly enhanced for ATP and UDP ( Fig.   5B, C).

Effect of pyrophosphate metabolism on drought tolerance
In order to evaluate the vitality of plants with reduced plastidial sPPase activity, a short term, mild drought stress was induced, and phenotypically and biochemically evaluated. Treatment with 10% (w/v) PEG-6000 (ψ ABA concentrations increased three-fold in the drought stressed control plants, while ABA levels of stressed psPPase plants were unaltered from the well-watered conditions (Fig. 6D). In addition, under drought induced conditions both GA 3 and IAA levels increased two-and three-fold, respectively, in the psPPase plants ( Fig.   6E, F). Other metabolic constituents such as, soluble sugar (glucose, fructose and sucrose) levels measured under drought stress were unaffected by the treatment (Table II). On the other hand, starch levels were decreased in psPPase stressed plants with respect to the stressed control (Table II). Also, the majority of the metabolite changes that were observed for the psPPase-silenced plants under well-watered conditions were also reflected when exposing the TRV2 control to drought stress. psPPase drought-stressed plants revealed few further changes compared to the TRV2 stressed control (Fig. 4, Table S2). The exception to this was a significant increase in 2-oxoglutarate levels (Fig. 4).

DISCUSSION
Over the past 20 years, PP i metabolism has been extensively studied in many species. In microorganisms and invertebrates it is known that soluble pyrophosphatase activity is necessary for growth and development (Chen et al., 1990;Pérez-Castiñeira et al., 2002;Islam et al., 2005;Ko et al., 2007). Knowledge about the role of pyrophosphate in plant metabolism, however, remains fragmented. Whilst several attempts have relied on addressing this by removal of PP i from a specific sub-cellular compartment, to our knowledge no study has been undertaken to examine the effect of PP i accumulation in a particular organelle. An The mottled appearance of light green islands along the leaf adaxial lamina of psPPase silenced plants (Fig. 2), suggested that isoprenoid biosynthesis could be compromised following PP i accumulation. In autotrophic metabolism isoprenoids act in fundamental roles as photosynthetic pigments (chlorophylls, carotenoids), electron carriers (quinones), radical scavengers (tocopherols), membrane components (sterols), as well as growth and defense regulators such as ABA, gibberellins (GAs), brassinosteroids, cytokinins, monoterpenes, sesquiterpenes, and diterpenes. Isoprenoids are synthesized from isopentenyl (IPP) and dimethylallyl diphosphate (DMAPP) precursors either in the cytosol via the mevalonate/acetate (MVA) pathway or through the plastidial-localized methylerythritol phosphate (MEP) pathway (Lichtenthaler, 1999; see Fig. S1).
Plastidial IPP pools have been previously shown to serve primarily as substrate for monoterpenes, diterpenes, tetraterpenes (carotenoids), prenyl moieties of chlorophyll, plastoquinone and tocopherol, as well as having a dual role with the MVA pathway to supply substrates for sesquiterterpene synthesis (Chappell, 2002;Dudareva et al., 2005). The action of the plastidial-localized phytoene synthase (Seo and Koshiba, 2002) is reliant on efficient PP i dissipation so its accumulation would be expected to affect synthesis of carotenoids, xanthophylls, ABA and GAs (see Fig. S1). In psPPase plants, carotenoid and the phenyl moieties investigated were largely, but not exclusively, down regulated (Table I). In the current study although violaxanthin was decreased, zeaxanthin was increased in the plants with lowered psPPase activity (Table I). Interestingly, both ABA and GA 3 levels remained invariable under unstressed conditions (see later for further discussion) despite the decreases in their precursor molecules.
Chlorophyll is also synthesized in a PP i -generating step via chlorophyll synthetase (see Fig. S1). Chl a, but not chl b amounts were significantly decreased compared to the controls (Table I). This led to a significant decrease in the chl a/b ratio (from 3.74 in control to 2.92 in psPPase), suggesting that the stoichiometry of the LHCs of PSII relative to PSI increased to compensate for the loss in pigment molecules. An alteration in the chl a/b ratio has also been suggested to affect retrograde signaling of photosynthetic genes (Pesaresi et al., 2007), however it is evident from our results that a post-transcriptional mechanism coordinates the accumulation of the Rubisco hetero-enzyme subunits (Fig. 3E). A plausible explanation for this is that RNA-polymerase and tRNA synthases (both of which produce PP i ; Fig. S1) are inhibited within the chloroplasts by the PP i concentrations in the silenced plants.
The reductions in photosynthetic pigments observed here as well as the reduced amounts of Rubisco suggested that carbon fixation may be affected in the psPPase silenced plants. Photosynthetic assimilation measurements confirmed that carbon fixation rate was reduced by approximately 55% under ambient CO 2 concentrations (Fig. 3A). The reduction in the photosynthetic rate in the silenced plants could be ruled out as related to phosphate limitation of photosynthesis due to the fact that the silenced plants do not contain less ATP than controls (Fig. 5B).
Reduced photosynthesis under saturating CO 2 concentrations suggests a further limitation in triose phosphate utilization (TPU) in the psPPase plants. This is supported by an inability of leaf discs to synthesize starch in dark-adapted, sucrose supplemented conditions, as well as enhanced 3-PGA levels observed (Fig. 3A,   Fig. 4). The reduction in carbon fixation is most likely caused by a combination of all of these factors.
As was noted above, the silenced plants in the current study were less able to synthesize starch than the controls (Table II, Fig. 3A). A likely explanation for these observations is that the increased PP i would affect ADP glucose pyrophosphorylase (AGPase). This enzyme catalyzes the first committed reaction in starch biosynthesis using glucose-1-phosphate and ATP to produce ADPglucose and PP i (Fig. S1) and is known to be essential for starch synthesis, Despite the increase in its activity in crude extracts, (Fig. 3B) increased plastidial PP i would be expected to make it's reaction less thermodynamically favourable in the forward direction in vivo (Amir and Cherry, 1972).
When grown under mild drought stress ABA concentrations were significantly decreased in the psPPase silenced plants compared to the controls ( Fig. 6D). ABA may be synthesized either through a 9-cis-violaxanthine (C40 indirect carotenoid) pathway (mainly found in higher plants) (Schwartz et al., 2003) or a farnesyl diphosphate (C15 precursor) pathway (Oritani and Kiyota, 2003).
Under drought conditions ABA is up regulated and triggers stomatal closure to limit water loss (Mittelheuser and van Steveninck, 1969). In the current study, psPPase repressed plants were characterized by reduced violaxanthin, normal neoxanthin and increased zeaxanthin content under prevailing greenhouse conditions (Table   I). ABA was also unchanged in the psPPase repressed plants under normal conditions and synthesis could not be induced when the plants were challenged with mild water stress. This suggested that stress-induced ABA synthesis in TRV2-psPPase plants are probably derived from violaxanthin. However, it cannot be ruled out that, under water stress conditions, the substrates might become limiting due to compensatory mechanisms whereby pigments are directed towards increasing the photoprotective capacities of the light harvesting complex (Snyder et al., 2006;Dall'Osto et al., 2007). In accordance with the ABA levels, no difference in transpiration rate between unstressed TRV2-psPPase and TRV2 control plants could be found under normal conditions. The rate decreased in both variants under drought stress; however, the decrease was far less severe in the TRV2-psPPase plants than in the TRV2 controls and led to the TRV2-psPPase plants wilting faster than the respective TRV2 control (Fig. 6A, B). In addition it could be speculated that the increased GA 3 and IAA levels in the silenced plants (Fig. 6E, F) contributes to the reduced ABA biosynthesis through alleviation of DELLA activation of a putative E3 ligase gene (Zentella et al., 2007) or other downstream targets.
In contrast to metabolites that are only evident when exposed to drought-stressed conditions, the TRV2-psPPase plants exhibited a drought mock response under well-watered conditions. These included the increases in metabolite levels of several phenylpropanoids or precursors (quinate, coniferylalcohol, ferulate, 3-caffeoyl quinate, 4-caffeoyl quinate and 5-caffeoyl quinate) which may be involved in lignification processes associated with drought stress (Lee et al., 2007). Protection against drought stress can also be facilitated by the induction of osmoprotective compounds such as the amino acids proline and glutamate, sugar or sugar polyols and/or inorganic ions (Mahajan and Tuteja, 2005, and references therein). Increases in 1-pyrroline-2-carboxylate (precursor to proline), myo-inositol and a metabolite similar to pinitol were also observed in this dataset (Fig. 4). When TRV2 control plants were drought induced, the majority of these metabolites accumulated in a similar manner to the TRV2-psPPase unstressed plant, and also were not significantly different from the TRV2-psPPase stressed metabolite levels (Fig. 4). This suggests that PP i metabolism might be indirectly involved in mediating drought stress responses in N. benthamiana leaves. Interestingly, also malate levels showed similar patterns to those observed for the osmoprotective responses. Whilst the exact mechanism remains unknown, modulation of malate levels in transgenic tomato leaves has been shown to induce opposing photosynthetic responses (Nunes-Nesi et al., 2005;2007), with antisense fumarase plants impaired in regulating stomatal aperture (Nunes-Nesi et al., 2007).
Cumulatively these results suggest that malate plays a profound role in mediating photosynthetic performance and that these responses may also be integrated with the prevailing PP i levels.
It remains unclear whether PP i is transported across the plastid membrane, and whether this could affect metabolism in different compartments. Lunn and Douce (1993) described a transporter from isolated chloroplast preparations that is able to import PP i over the chloroplast membrane, however, neither a PP i export mechanism nor the corresponding gene have been isolated to date. Similarly, in developing maize embryos L-malate/PP i transport has been demonstrated (Lara-Núñez and Rodríguez-Sotres, 2004) but gene identification remains elusive.
In summary, examining the effect on metabolism of a repression of plastidial pyrophosphatase has demonstrated the essential role that PP i plays in many plastidial pathways. Increased PP i concentrations led to reduced accumulation of several chloroplastidial localized metabolites that are important for plant survival, such as ABA, chlorophyll and carotenoids. This indicated that repression of psPPase is extremely detrimental to the plant as photosynthesis is reduced and the plant became unable to regulate its water exchanges under mild drought stress due to an inability to manufacture ABA. Taken together these data indicate that psPPase plays an extremely important role in plastidial metabolism and, similar to microorganisms (Chen et al., 1990;Pérez-Castiñeira et al., 2002), we would suggest that a mutation eliminating this soluble pyrophosphatase would be lethal.

VIGS plasmids for psPPase transient repression
The tomato EST clone cLET20N17, obtained from the Clemson University Genomics Institute, was digested with KpnI and BamHI and ligated into the same restriction sites in the tobacco mosaic rattle virus vector pTRV2. Deconstructed vectors (pTRV1, pTRV2, and pTRV2-PDS) were transformed into A. tumefaciens At the four leaf seedling stage, plants were transferred to 1 L tissue culture containers with the same growth media constituents for ten days before being subjected to VIGS infiltration. Seedlings (4-5 weeks after germination) were vacuum infiltrated with 20 ml of transformed Agrobacterium suspension (containing a 1:1 mix of TRV1 and either TRV2 or TRV2-psPP). The air volume was adjusted to 20 ml before the nib of the syringe was stoppered. While the leaves of the plant were submerged, the air volume was increased to 40 ml, corresponding to a vacuum of 50 kPa, and held there for 30 seconds. Infiltrated seedlings were planted and grown in sterile potting soil (with silica and vermiculite [8:1:1]) during the summer months under prevailing greenhouse conditions without any additional carbon supplementation. After three weeks, leaves were harvested at midday (unless stated otherwise), immediately frozen in liquid nitrogen and stored at -80 o C until further use. Samples were either whole leaves (leaf number three to five), or 64 mm 2 leaf discs taken from leaf three using a cork borer, and homogenized prior to sample processing. In order to monitor transfection efficiency and growth conditions, the TRV2-PDS vector (containing a cDNA encoding phytoene desaturase; Liu et al., 2002b) was used in parallel as an internal in-house control.
All of the phenotypic alterations documented in this paper were observed in at least three independent infiltration experiments.

Plastidial isolations and enrichment determinations
Leaf material was harvested from plants which had been destarched by darkening over a 48 h period. Enrichments were then performed according to (Kubis et al., 2008). Plastids were collected and centrifuged at 3000g for 2 min, 4°C, the pellet was re-suspended in 500 µl protein extraction buffer described above and sonicated for three 1 second bursts separated by 10 second incubations on ice.

In-gel sPPase activity and denaturing SDS PAGE analyses
sPPase in-gel assays were performed by running either total or plastidial enriched protein extracts on a non-denaturing 10% (w/v) polyacrylamide gel at 4°C.
The gel was incubated in 20 ml pyrophosphatase assay buffer described in

Enzyme assays
In gel assays for PGM were performed by loading 30 µg of total protein onto a 10% (v/v) native polyacrylamide gel containing 4 mg glycogen (Mytilus edulis type VII, Sigma Co., MO, USA). The gel was stained according to Vallejos (1983).

Semi-quantitative RT-PCR analysis
Total RNA was isolated from 200 mg leaf material using the phenolchloroform method. First strand cDNA synthesis using oligodT primers was

Stomatal conductance was measured on a EGM-4 Environmental Gas
Monitor (PP Systems, USA) Readings were taken directly prior to harvesting at midday on the fourth leaf of each plant with the flow rate (50 ml.min -1 ) and temperature (25 o C) kept constant. Carbon assimilation rates were analyzed using an infrared gas analyzer (Ciras-1, PP system, UK) on the third fully expanded leaf from psPPase and control plants at 0, 380, and 1000 µmol.mol -1 intercellular CO 2 concentrations with a flow rate of 350 ml.min -1 , 25°C and constant PAR of 1400 μ mol.m -2 .s -1 .

Metabolite determinations
PP i was extracted from leaf tissue by the TCA/ether method (Jelitto et al., 1992). PP i was determined using the colorimetric PiPer pyrophosphate cycling assay kit (Invitrogen, CA, USA) according to the manufacturers' specifications. All porcelain and glassware was pre-treated overnight with 0.1 M HCl to remove residual phosphate. PP i levels were determined by a sample blank with or without sPPase, and total P i calculated by comparison of fluorescence at 595 nm with a linear P i standard curve.
Soluble sugars and starch were extracted and assayed according to Müller-Röber et al. (1992).
Primary metabolites levels were extracted and derivatized as previously Adenylates and uridinylates were extracted and detected as described by Fernie et al. (2001).
Carotenoids and xanthophylls was extracted and determined according to Taylor et al. (2006). Homogenized leaf discs were incubated for 5 min with 100 µl methanol containing β -apo-caroten-8-al as internal standard, 100 µl 50 mM Tris-HCl (pH 8.0) and 1 M NaCl added and incubated for another 5 min. The mixture were partitioned twice with 400 µl chloroform, centrifuged at 3000g for 5 min, 4°C, the lower phases pooled and dried under vacuum. Samples were immediately resuspended in ethyl acetate:methanol (1:4) with 0.1% (w/v) butylated hydroxytoluene (BHT) and ran according to Taylor et al. (2006). The peak area was integrated and normalized with respect to 53.5 ng β -apocaroten-8-al injected.

Phytohormone profiling
Phytohormones were extracted according to Edlund et al. (1995). In brief, 500 µl of 0.05 M Na-phosphate buffer (pH 7.0) was added in a 10:1 ratio to 1 homogenized leaf tissue, and incubated for 1h in the dark, continuous shaking, 4 o C. After extraction, the pH was adjusted to 2.6, the sample enriched with ca. 35 mg Amberlite XAD-7 (Serva, Heidelberg, Germany) and further incubated for 1 h in the dark, continuous shaking at 4 o C. After centrifugation, the XAD-7 was washed twice with 500 µl 1% (v/v) acetic acid before elution with 500 µl dichloromethane Running conditions was exactly as previously described (Edlund et al., 1995), and phytohormone identification and quantification was done by means of linear calibration curves for authentic standards and the mass spectrums adjusted accordingly.

Statistical analysis
Unless otherwise specified, statistical analyses were performed using Students t-test embedded in the Microsoft Excel software (Microsoft, Seattle). Only the return of a P value < 0.05 was designated significant. Analysis of variance followed by Fisher's least significant difference test was conducted either in