Silencing of the mitochondrial ascorbate synthesizing enzyme L-galactono-1,4-lactone dehydrogenase affects plant and fruit development in tomato.

l-Galactono-1,4-lactone dehydrogenase (EC 1.3.2.3) catalyzes the last step in the main pathway of vitamin C (l-ascorbic acid) biosynthesis in higher plants. In this study, we first characterized the spatial and temporal expression of SlGalLDH in several organs of tomato (Solanum lycopersicum) plants in parallel with the ascorbate content. P35S:SlgalldhRNAi silenced transgenic tomato lines were then generated using an RNAi strategy to evaluate the effect of any resulting modification of the ascorbate pool on plant and fruit development. In all P35S:SlgalldhRNAi plants with reduced SlGalLDH transcript and activity, plant growth rate was decreased. Plants displaying the most severe effects (dwarf plants with no fruit) were excluded from further analysis. The most affected lines studied exhibited up to an 80% reduction in SlGalLDH activity and showed a strong reduction in leaf and fruit size, mainly as a consequence of reduced cell expansion. This was accompanied by significant changes in mitochondrial function and altered ascorbate redox state despite the fact that the total ascorbate content remained unchanged. By using a combination of transcriptomic and metabolomic approaches, we further demonstrated that several primary, like the tricarboxylic acid cycle, as well as secondary metabolic pathways related to stress response were modified in leaves and fruit of P35S:SlgalldhRNAi plants. When taken together, this work confirms the complexity of ascorbate regulation and its link with plant metabolism. Moreover, it strongly suggests that, in addition to ascorbate synthesis, GalLDH could play an important role in the regulation of cell growth-related processes in plants.

L-Galactono-1,4-lactone dehydrogenase (EC 1.3.2.3) catalyzes the last step in the main pathway of vitamin C (L-ascorbic acid) biosynthesis in higher plants. In this study, we first characterized the spatial and temporal expression of SlGalLDH in several organs of tomato (Solanum lycopersicum) plants in parallel with the ascorbate content. P 35S :Slgalldh RNAi silenced transgenic tomato lines were then generated using an RNAi strategy to evaluate the effect of any resulting modification of the ascorbate pool on plant and fruit development. In all P 35S :Slgalldh RNAi plants with reduced SlGalLDH transcript and activity, plant growth rate was decreased. Plants displaying the most severe effects (dwarf plants with no fruit) were excluded from further analysis. The most affected lines studied exhibited up to an 80% reduction in SlGalLDH activity and showed a strong reduction in leaf and fruit size, mainly as a consequence of reduced cell expansion. This was accompanied by significant changes in mitochondrial function and altered ascorbate redox state despite the fact that the total ascorbate content remained unchanged. By using a combination of transcriptomic and metabolomic approaches, we further demonstrated that several primary, like the tricarboxylic acid cycle, as well as secondary metabolic pathways related to stress response were modified in leaves and fruit of P 35S :Slgalldh RNAi plants. When taken together, this work confirms the complexity of ascorbate regulation and its link with plant metabolism. Moreover, it strongly suggests that, in addition to ascorbate synthesis, GalLDH could play an important role in the regulation of cell growth-related processes in plants.
Vitamin C (L-ascorbic acid) has a huge importance for all living eukaryotic cells. In higher plants, ascorbate is the most abundant water-soluble antioxidant, acting to scavenge reactive oxygen species that are generated during photosynthesis, oxidative metabolism, and a wide range of stresses. In addition, ascorbate is involved in the regulation of several fundamental cellular processes such as photoprotection, the cell cycle, cell expansion, and in pathways of secondary metabolism such as the recycling of lipid-soluble a-tocopherol and ethylene biosynthesis. Although generally accepted, the mechanism by which ascorbate participates in the cell developmental processes is not clearly established (for review, see Smirnoff, 2000). Much evidence supports the correlation between ascorbate content and its redox state and cell division and/or cell expansion (Horemans et al., 2003 and refs. therein).
Microorganisms, plants, and most animals produce ascorbate, however, a loss of the ability to synthesize ascorbate has occurred in primates and some other animals. Since the 1960s, the animal pathway has been completely characterized (Burns, 1960). It involves D-Glc as the initial precursor and the last step is catalyzed by a microsomal L-gulono-1,4-lactone oxidase, which oxidizes L-gulono-1,4-lactone to produce ascorbate. In plants, the ascorbate pathway is different from animals and has remained unsolved until recently, when convincing evidence in support of a novel pathway was established (Wheeler et al., 1998). To date, the pathway proposed by Wheeler et al. (1998) is the most commonly described in plants. Recently the last missing intermediate steps of the pathway have been elucidated (Laing et al., 2007;Linster et al., 2007). The last step in this pathway clearly involves the conversion of L-galactono-1,4-lactone (L-GalL) to ascorbic acid, a reaction catalyzed by the L-galactono-1,4lactone dehydrogenase (here referred to as L-GalLDH), an enzyme located in the inner membrane of the mitochondria (Bartoli et al., 2000). This reaction requires cytochrome c as a second substrate that is reduced as L-GalL is converted into ascorbate (Ô ba et al., 1995).
In addition to this pathway, alternative ascorbate biosynthesis pathways have been proposed in plants. Agius et al. (2003) showed that during the ripening phase of strawberry (Fragaria spp.) fruit, ascorbate is produced via the reduction of D-GalUA to L-galactonic acid, which in turn is spontaneously converted to L-galactono-lactone, the immediate precursor of ascorbate. However, this pathway shares the requirement of L-GalLDH activity as the terminal step in ascorbate synthesis. In 2003, Wolucka and Van Montagu suggested an alternative pathway involving L-gulose as an intermediate for the de novo biosynthesis of vitamin C in plants. They showed that in vitro, GDP-D-Man-3,5epimerase is capable of catalyzing two distinct epimerization reactions that produce either GDP-L-Gal or the novel intermediate, GDP-L-gulose. Nevertheless, the existence of all the enzyme machinery of this putative pathway has yet to be confirmed in vivo. Recently, Lorence et al. (2004) presented molecular and biochemical evidence proposing another possible biosynthetic route that utilizes myoinositol as a precursor. By overexpressing myoinositol oxygenase in Arabidopsis (Arabidopsis thaliana), they observed a 2-to 3-fold increase in ascorbate content in the leaf and discuss the possible contribution of myoinositol in ascorbate biosynthesis in plants. Following constitutive expression of the rat gene encoding L-gulono-lactone (L-GulL) oxidase, Jain and Nessler (2000) achieved a 4-and 7-fold increase of ascorbate content in lettuce (Lactuca sativa) and tobacco (Nicotiana tabacum) plants, respectively. L-GulL oxidizing activity has been reported in Arabidopsis by Davey et al. (1999) who measured L-GulL oxidizing activity and by Wolucka and Van Montagu (2003) who measured L-GulLDHlike activity.
Ascorbate is known to be a vital physiological metabolite in plants for major plant biochemical processes such as photosynthesis (Smirnoff, 1996;Noctor and Foyer, 1998), but also for controlling plant development as recently shown in root and other organs (Olmos et al., 2006). To investigate the physiological function of L-GalLDH in plants, we analyzed the effect of L-GalLDH silencing on ascorbate metabolism and its consequences for plant development. Since fruit is one of the major sources of ascorbate for the human diet, this study was carried out in tomato (Solanum lycopersicum), the model plant for fleshy fruit development. We report here that SlGalLDH silencing profoundly affects plant and fruit growth, probably through the alteration of the mitochondrial function and related changes in ascorbate redox state, particularly in the apoplast. Transcriptome and metabolome analyses of P 35S :Slgalldh RNAi transgenic lines further suggest that phenotypic changes can be attributed to the resulting modifications in secondary and primary metabolism.

Ascorbate and L-GalLDH Distribution in Tomato Plant
To extend our understanding of the role of ascorbate in plant growth, we analyzed its distribution in a wide range of tomato organs, including vegetative tissues such as young and mature leaf, stem and root, and reproductive tissues such as flower and fruit at several stages of development (Fig. 1, A and B). In vegetative Figure 1. SlGalLDH expression and ascorbate content in cherry tomato plants and fruit. A, Relative SlGalLDH transcript levels in young leaves (Yl), mature leaves (Ml), root (Rt), stem (St), flower (Fl) and in fruit at 10 DPA, 20 DPA, mature green (MG), orange (Or), and red ripe (RR) stages. Data obtained by semiquantitative RT-PCR were normalized against Actin1 mRNA and are expressed as a ratio of arbitrary units. B, Total ascorbate content in the various tomato organs. Data represent mean 6 SD of measurements of 10 organs per plant with six individual plants per line (n 5 60). C, Detection of SlGalLDH transcripts in developing tomato organs by in situ hybridization. Longitudinal sections of shoot apical meristem (A), root apical meristem (B), young leaf (C), 9 mm flower bud (D), and cross section of fruit at 7 DPA (E) and 20 DPA (F) were prepared and analyzed as described in the experimental procedures. Hybridization signal appears as dark staining. Inserts are negative control corresponding to sense riboprobe. Scale bar 5 500 mm. organs, ascorbate content varied according to the tissue, with high levels in the leaf (young and mature) and comparatively low levels in the root. In flower and fruit, the ascorbate content remained very low (1.7-2.3 mmol per g fresh weight [FW]) compared to the leaf (6.8-7.9 mmol per g FW). The ascorbate concentration was constant throughout fruit development and comparable with the levels reported by Andrews et al. (2004).
Since L-GalLDH catalyzes the terminal step of the major pathway of ascorbate biosynthesis in plants (Wheeler et al., 1998), this enzyme is a good candidate for controlling the variations in ascorbate content in plants (Tamaoki et al., 2003;Bartoli et al., 2005;Tokunaga et al., 2005). Screening of tomato EST databases (SOL Genomics Network [SGN] and The Institute for Genomic Research tomato EST databases; http:// www.sgn.cornell.edu and http://www.tigr.org/) allowed the identification of a single cDNA encoding L-GalLDH (named SlGalLDH for tomato L-GalLDH), which corresponds to a unique gene in tomato, as further confirmed by Southern-blot analysis (data not shown). Examination of SlGalLDH expression showed that SlGalLDH transcripts (Fig. 1A) and immunodetected SlGalLDH protein (data not shown) were ubiquitously detected in the various tomato organs analyzed. The high SlGalLDH mRNA abundance in the leaf was in agreement with the high ascorbate content in this tissue (Fig. 1B), while other organs such as root or fruit did not display this relationship. Close examination of the spatial distribution of SlGalLDH transcripts in developing organs by mRNA in situ hybridization ( Fig.  1C) further indicated that SlGalLDH transcripts were particularly abundant in tissues displaying a high cell proliferation activity such as the shoot apical meristem, the root apical meristem, young leaves, ovules, and stamen of flowers and young fruit at 10 DPA. To further investigate the relationship between plant development and SlGalLDH activity in tomato, we generated transgenic plants expressing a SlGalLDHspecific RNA interference sequence fragment under the control of the cauliflower mosaic virus 35S promoter (hence named P 35S :Slgalldh RNAi transformants). Among 12 independent primary P 35S :Slgalldh RNAi transformants showing the presence of a single copy of the transgene, we selected four transgenic lines showing a range of phenotypic changes varying from severe (line 8) to moderate (line 5) and slight (lines 38 and 2) for further analyses on the next generation (T1 plants). To verify that the morphological changes observed in primary transformants were linked to the RNAimediated reduction in SlGalLDH mRNA, the detailed analysis of SlGalLDH mRNA and protein abundance and activity was performed in leaf and/or 20 and 42 DPA fruit from the corresponding T1 homozygous P 35S :Slgalldh RNAi transgenic lines 2, 5, 8, and 38 and from controls. In all transgenic lines, SlGalLDH mRNA, protein abundance, and residual SlGalLDH activity were highly reduced by comparison with the control plants (Fig. 2), and no significant difference could distinguish the four transgenic lines on the basis of the expression/protein/activity data when statistical analysis was performed. Despite these strong effects, no remarkable variation could be observed in the total ascorbate content in young fully expanded leaves (6.75 6 0.04 mmol per g FW) and 45 DPA red ripe fruits (1.82 6 0.03 mmol per g FW) in the lines 2, 5, 8, and 38. The capacity of the transgenic plants to maintain the pool of total ascorbate constant was confirmed for the leaves of the lines 5 and 8 (Fig. 3). Leaves from the transgenic lines and controls accumulate ascorbate at similar rates when incubated with the ascorbate precursor L-GalL. Further investigations indicated that no viable plant with SlGalLDH activity below the threshold level of 15% was ever recovered among the P 35S :Slgalldh RNAi transformants, suggesting the essentiality of this activity. Given the well established fact that ascorbate content of the leaf and ascorbate redox state depend on many environmental parameters, P 35S :Slgalldh RNAi transgenic lines 5 and 8 and controls were grown in a phytotronic chamber to avoid any daily climate changes. Measurements of the total ascorbate content (reduced ascorbate [AsA] 1 dehydroascorbate) as well as of the AsA content were carried out during the development of these plants, from the seed up to the appearance of the first inflorescence (6 weeks after sowing), and within the developing fruit (Table I). Total ascorbate content in the seed and the young leaves of the apex was comparable in both transgenic and control plants. In the germinating seed of transgenic lines and the controls, approximately 90% of ascorbate was as the reduced form AsA. The leaf ratio of AsA to total ascorbate declined, both in the controls and the transgenic lines, throughout plant development (Table I). This effect, which was already noticeable in 21 d after sowing (DAS) transgenic plants, was not due to variations in total ascorbate content but to  Table I. Ascorbate in tomato leaf and fruit of P 35S :Slgalldh RNAi transgenic and control plants Reduced (AsA) and total ascorbate (Tot Asc) contents of P 35S :Slgalldh RNAi line 8, line 5, and control plants were measured in young developing leaves from seedlings at the indicated DAS and in developing fruits at the indicated days postanthesis (DPA). Apoplastic ascorbate content was measured in the fully expanded fourth leaf of 6-week-old plants. The reduced/total ascorbate ratio was determined for all the samples. Data represent mean 6 SD of six individual plants with two repeats per plant. The superscript a ( a ) indicates values that were determined by the t test to be significantly different (P , 0.05) from control. variations in AsA content. The total ascorbate content of control fruit grown in a controlled environment chamber (Table I) was similar to that observed in control plants in the greenhouse (Fig. 1B). In the control fruit, 80% of total ascorbate was in the form of oxidized ascorbate during the cell division and expansion phases (10-20 DPA), while more than 90% of the total ascorbate was in the form of AsA in the ripening fruit. Total ascorbate content from transgenic lines 5 and 8 fruit were not significantly different from fruit of the control, except for the 10 DPA stage that exhibited a slightly higher ascorbate content. In contrast with the leaf, transgenic fruit from lines 5 and 8 exhibited a significant increase in AsA (Table I), mainly during the cell division and cell expansion stages of the fruit (10-30 DPA). Because of the putative role of apoplastic ascorbate in the signaling processes modulating cell growth and development (Pignocchi and Foyer, 2003), we investigated the total and reduced ascorbate content in the apoplast of fully expanded leaves of 6-week-old P 35S :Slgalldh RNAi plants of transgenic lines 8 and 5. In the leaves of control plants, only 10% of the total ascorbate was found in the apoplast (648 6 49 nmol per g FW) in agreement with previous findings of Noctor and Foyer (1998), and 10% to 15% of the apoplastic ascorbate was present as the reduced form. In the P 35S :Slgalldh RNAi transgenic line 8 and line 5, the reduced ascorbate pool remained very low (85-121 nmol per g FW), as in the controls. This occurred despite the fact that the apoplastic total ascorbate content was decreased by about 50%, due to a reduction of the oxidized ascorbate pool (Table I).
Given that Millar et al. (2003) demonstrated that GalLDH activity is associated with the complex I of the mitochondrial electron transport chain and that it has been documented that respiration can control ascorbate synthesis in plants ( Bartoli et al., 2006), intact leaf mitochondria were isolated to investigate respiratory parameters in the P 35S :Slgalldh RNAi transgenics. When tricarboxylic acid (TCA) cycle substrates pyruvate and malate were used to drive the electron chain, as well as when electrons entered the chain via succinate dehydrogenase by the addition of succinate, the respiration rate in the transgenic line 5 and 8 was identical to the controls (Table II). However, when external NADH dehydrogenases were engaged by addition of exogenous NADH, the respiration rate in P 35S :Slgalldh RNAi transgenic line 8 was increased by 1.6-fold whereas that in P 35S : Slgalldh RNAi transgenic line 5 was similar to the rate of control plant mitochondria. In the same order, a significant 2-fold activation of cyanide-insensitive (alternative oxidase) respiration that was inhibited by salicylhydroxamic acid (SHAM), was measured in mitochondria from P 35S :Slgalldh RNAi line 8. When no substrate, other than L-GalL, was used to introduce electrons flow to cytochrome c, the respiration rate in both transgenic lines was reduced to about 50% of that observed in mitochondria isolated from the control. Table II. Respiration and SlGalLDH activity in isolated mitochondria from leaves Respiration was measured in intact mitochondria prepared from young developing leaves of P 35S :Slgalldh RNAi line 8, line 5, and control plants as described in the ''Materials and Methods.'' The reaction was conducted with 500 mg of mitochondria protein in 1 mL of respiration buffer in the presence of substrates at concentration as indicated. Prior to measuring the respiration in the presence of L-GalL, the mitochondria were incubated with 12.5 mM antimycine A. SlGalLDH activity was measured with 30 mg of mitochondrial protein as described in the ''Materials and Methods.'' Data represent mean 6 SD of seven or eight individual mitochondrial isolations. Superscript a ( a ) indicates values that were determined by the t test to be significantly different (P , 0.01) from control. The difference in the respiratory activities in the different transgenic lines demonstrates that a threshold level of 50% must be exceeded for the reduction of GalLDH abundance to impact the mitochondrial electron transport chain.
Transgenic Plants with Reduced L-GalLDH Activity Display a Reduced Plant Growth Rate and Fruit Size When T1 transgenic plants were grown in the greenhouse, it soon became apparent that the growth of the aerial parts was reduced compared with that of the controls. Detailed characterization of this effect revealed that the germination rate, the plant growth rate, the fruit weight, and the diameter were all reduced (Fig. 4). When this experiment was replicated in a growth chamber with tightly controlled growth conditions similar results were obtained. To investigate how the reduction in SlGalLDH mRNA abundance, protein, and activity could affect plant growth and morphology in the transgenic plants, we conducted more detailed analyses of lines 2, 5, 8, and 38. Given the reduction in aerial yield in the P 35S :Slgalldh RNAi transformants and the importance of ascorbate for the photosynthetic process, we next assessed to what extent the transgenic plants exhibited altered photosynthetic rates. The rate of CO 2 assimilation was measured on fully expanded leaves of the transformant and control plants at incident irradiance of between 200 to 1,000 mmol m 22 s 21 . At this light intensity range, the CO 2 assimilation curve was linear and the calculated photosynthetic rate, 16.2 6 0.8 nmol CO 2 per mmol photon, was invariant in transformants with respect to the controls. It is noteworthy that chlorophyll content was also unaltered (1.4 6 0.2 mg chl per g FW).
In contrast, leaf size was strongly affected in the P 35S :Slgalldh RNAi transgenics, which resulted from a 25% reduction in leaflet area in the most strongly inhibited lines (Fig. 5, A and B). Analysis of epidermal cell size indicated that this result does not stem from a reduction in cell number but rather from a reduction in cell size, which was also reduced by about 25% compared to that of the control (Fig. 5B). Similar observations were made in fruit, which exhibited a cell size reduction of 15% to 22% in the transgenic lines 8 and 5, despite exhibiting an unaltered ripening period (45 d). The analysis of a cross section taken from the equatorial area of a red ripe fruit revealed that the transgenic lines produced fruits with a thinner pericarp. Given that it is very difficult to obtain suitable samples from red ripe fruit for histological analysis (Cheniclet et al., personal communication), we compared the thickness of the pericarp in fruits at 20 DPA, at which stage the fruits reached 85% of their final size. Cytological analysis revealed that the fruit pericarp thickness was significantly decreased in all lines (Fig. 5C). Detailed comparisons indicated that this was a result of a cell size reduction in the pericarp rather than an alteration in the number of cell layers (Fig. 5D).  SlGalLDH activity and/or variations in ascorbate redox state to the morphological and cellular alterations of transgenic plants (Figs. 4 and 5), we compared the transcript expression profiles of a fully expanded fourth leaf and of 42 DPA (orange) fruit in the severely affected line 8 and in control plants. Among the 13,400 cDNAs corresponding to 8,700 transcripts present on the TOM1 tomato microarray analyzed, 1,269 in the leaf and 92 in the fruit were significantly different (P value , 0.02, four independent slide hybridizations, including two dye swaps and two replicates per slide) in line 8 versus control plants (see http://cbi.labri.fr/ outils/data/Tomato/VitC/sup.html). A previously reported and recently further enhanced annotation of the genes represented on the TOM1 array combined with the MapMan ontology (Thimm et al., 2004;Urbanczyk-Wochniak et al., 2006) was used for analyses (see http://cbi.labri.fr/outils/data/Tomato/VitC/sup. html). More detailed analyses were performed on the most differentially expressed genes selected using a mean ratio threshold .1.6, representing 35 genes in the fruit and 83 genes in the leaf; of these, seven genes in the fruit and 19 in the leaf encoded proteins with unknown functions or, alternatively, presented no homology with known genes (Table III). Most of the genes identified were up-regulated in the leaf from the transgenic lines, only 5% of the differentially expressed genes were repressed in either leaf or fruit.
Classification of the known genes into the different functional categories defined by the Munich Information Center for Protein Sequences (http://mips.gsf.de/ projects/funcat) indicated the following distribution in leaf and fruit tissues: In the leaf, a first group (I) contained 18 stress-related transcripts known to be induced by various stresses (wounding, cold, light, etc.), such as geranylgeranyl reductase (Giannino et al., 2004) and Glu decarboxylase (Bouché and Fromm, 2004). A second group (II) contained 21 sugar metabolism-related genes involved in photosynthesis, carbohydrate, and cell wall metabolism, e.g. the Rubisco activase, the xyloglucan endotransglucosylase hydrolase, the PSII (or PSI) proteins, and the glyceraldehyde-3-P dehydrogenase. A third group (III) of 10 transcripts included several genes possibly involved in regulatory processes, for example a protein phosphatase 2C (PP2C) phosphatase, a MADS-box protein TDR3, and an F-box protein. A last group (IV) contained 10 genes encoding miscellaneous proteins related to transport and various metabolisms.
In the fruit, group I contained six stress-related transcripts including three heat shock proteins. Group II contained six transcripts, two of them encoding a PSII protein and Rubisco. Group III contained 12 transcripts, including a Ser/Thr protein kinase, a PP2C phosphatase, a WRKY type, and the agamous TAG1 transcription factors and a GA 20-oxidase. Remarkably, the MULTIPROTEIN BRIDGING FACTOR1 (MBF1) gene, a transcriptional coactivator known to induce the expression of stress-related genes involved in plant defense (Suzuki et al., 2005) and expressed in tomato (Zegzouti et al., 1999), displayed a high expression in the transgenic line 8 (ratio .4.5) compared to control plants. In addition, several genes involved in ethylene signal transduction such as the ethylene receptor ETR2 or in ethylene response, including the MBF1 factor, the ethylene-responsive protein ERF1, and the ripening-regulated protein DDTFR8 (Zegzouti et al., 1999) were strongly induced in the ripening fruit. Group IV contained four genes encoding proteins related to miscellaneous metabolisms.  We next analyzed the relationship between the phenotypes of three P 35S :Slgalldh RNAi lines (2, 8, and 38) and the changes in the redox state of ascorbate via a metabolomic approach. We analyzed the major pathways of primary plant metabolism by using an established gas chromatography-mass spectroscopy method (Fernie et al., 2004). As would be expected, the metab-olomic analyses confirmed the rank order of the lines, 5 and 8 being the most affected at the metabolic level (Supplemental Table S2). Furthermore, expanding leaves and orange fruits behaved inversely at the metabolic level ( Fig. 6; Supplemental Table S2). In leaves, metabolite analyses revealed a decrease in the levels of the major amino acids whereas some of them, Cys, Pro, Thr, and Val increased. In contrast, in orange fruit, the levels of amino acids were largely unaffected, with the exception of Val and Met that increased and of Pro that was reduced by half, significantly in the case of line 8 (Fig. 6). In leaves, the level of TCA cycle intermediates was significantly reduced, with the exception of succinate that displayed a slight but significant increase (Fig. 6). Among other organic acids, threonic acid, which corresponds to a degradation product of the ascorbate turnover pathway, displayed an increase in all transformants, whereas the opposite was observed for the g-aminobutyric acid (GABA) content. In orange fruits, the content of the TCA cycle intermediates was unchanged; however, the succinate content displayed an increase similar to that observed in the leaf. The most important change was for malate, which increased by up to 4.6-fold in line 8. Among the other organic acids, GABA accumulated up to 2.8-fold in fruit of line 8 whereas threonate was slightly reduced. Analysis of the leaf carbohydrate content revealed that the P 35S :Slgalldh RNAi transgenic lines were characterized by increases in Fru and Suc, as well as other sugars linked to the cell wall metabolism like Ara, Gal, Man, Rha, and raffinose, which increased by up to 5-fold in line 8. Myoinositol increased significantly in both SlGalLDH lines. In fruit, the level of sugars was largely unaffected; however, the levels of some cell wall-related sugars and myoinositol were significantly reduced, whereas sorbitol and mannitol increased. Interestingly, GalUA, an intermediate of the alternative ascorbate biosynthesis pathway and a degradation product of the cell wall pectins during fruit ripening (Agius et al., 2003) decreased significantly in all transgenic plants. Among the other compounds analyzed, it is interesting to note that intermediates involved in membrane biogenesis like C16:0 and C18:0 fatty acids were increased (Supplemental Table S2).

DISCUSSION
The RNAi-reduced expression of SlGalLDH in tomato leads to plants with a residual L-GalLDH activity showing several defects in growth rate and organ size (Figs. 4 and 5). All these observations are consistent with the preferential localization of SlGalLDH transcripts in tissues actively engaged in cell division and/ or expansion such as root tips or floral meristem and young fruit, as well as the photosynthetic leaves (Fig. 1). Among the primary transformants, two P 35S :Slgalldh RNAi transgenic lines showed the most extreme reduction in total ascorbate content (85-170 nmol ascorbate per g FW) and displayed very severe growth defects such as stunted plants with deformed leaves. These plants remained unable to set flowers and fruits and could not be propagated by cuttings, which prevented their use in subsequent studies. This result suggests the existence of a threshold level of L-GalLDH activity, below which ascorbate content and consequently plant growth are so severely impaired that reduction in L-GalLDH activity can be lethal to the plant. Another intriguing result to emphasize is the apparent complexity of ascorbate regulation in different plant tissues. Silencing of GalLDH had opposite effects in leaves and fruits on metabolic profiles and ascorbate redox state (Supplemental Tables S1 and S2). This discrepancy may result from the different functions of these organs, source (leaves), or sink (fruit). Most probably, it may result from the close relationship between photosynthetic electron transport in chloroplasts and ascorbate pool size in leaves, suggested by Yabuta et al. (2007). Silencing GalLDH has a significant effect on both the transcriptome and metabolome, thus giving emphasis to the importance of the role of GalLDH in plants, and data presented here allow us to draw the following conclusions.

SlGalLDH Activity Is Essential for Plant and Fruit Growth in Tomato
How the plant compensates for SlGalLDH reduction in the P 35S :Slgalldh RNAi transgenic lines studied (Table I; Fig. 2), to maintain a pool of ascorbate similar to the controls, remains an open question. The most plausible explanation is that residual SlGalLDH activity is high enough in the transgenic lines we tested to sustain sufficient ascorbate biosynthesis. This is supported by the experiment of incubation with L-GalL (Fig. 3) showing that even in lines 5 and 8, in which the activity was severely affected, the ascorbate accumulation rate was maintained at the same level as that of control plants. However, we cannot rule out that the maintenance of the ascorbate pool is associated with a reduction of its turnover. The metabolic analysis of fruit from P 35S :Slgalldh RNAi lines gives some support to this hypothesis in that a significant decline in the content of threonate (Fig. 5), the end product of one of the known pathways of ascorbate degradation (Green and Fry, 2005;DeBolt et al., 2006), exists. However, it should be noted that in leaves from the transgenic plants, threonate and its precursor dehydroascorbate strongly accumulated ( Fig. 6; Supplemental Table S2), raising the possibility of different regulation of ascorbate degradation in vegetative and reproductive organs. Another possible way for these plants to maintain the vital pool of ascorbate is the activation of alternative pathways of ascorbate biosynthesis (Agius et al., 2003;Wolucka and Van Montagu, 2003;Lorence et al., 2004). The pathway described by Agius et al. (2003) also requires L-GalLDH to catalyze ascorbate biosynthesis. While the gulose pathway proposed by Wolucka and Van Montagu (2003) still remains hypothetical, another plant pathway involving myoinositol has been proposed by Lorence et al. (2004). Elevated myoinositol levels are usually associated with stress conditions (Nelson et al., 1998). In P 35S :Slgalldh RNAi lines, the significant increase in myoinositol level in the leaf (Fig. 6; Supplemental Table S2) could, thus, be indicative of the function of this alternative pathway in vegetative organs.

Reduced SlGalLDH Expression May Affect Cell Growth by Modifying Mitochondrial-Related Energy Metabolism
Given the accepted role of ascorbate in photosynthesis, the hypothesis of an alteration in the photosynthetic capacity could have been anticipated in the P 35S :Slgalldh RNAi lines. However, our results clearly indicate that the reduced organ growth in the P 35S :Slgalldh RNAi lines does not result from an impaired photosynthesis, which was not affected in these plants despite the fact that more than 10 of the significantly up-regulated transcripts were directly related to photosynthesis (Table III).
One of the most important metabolic consequences that could explain cell growth impairment in the leaf is the change of mitochondrial function, as shown by the alteration of the Kreb's cycle (Fig. 6) and respiration rate observed in the most affected line 8 (Table II). This finding is interesting regarding the results of a recent study that supports a strong functional link between respiration and ascorbate synthesis and suggests that L-GalLDH is associated with complex I of the mitochondrial electron transport chain (Millar et al., 2003). This functional association may explain why the capacity of L-GalLDH is variable dependent on growth light intensity (Smirnoff, 2000;Bartoli et al., 2006) since the activity of the mitochondrial electron transport chain also varies with light intensity (Raghavendra and Padmasree, 2003). It is thus conceivable that the removal of most of the L-GalLDH protein from complex I affects electron transport. In support of this theory, levels of organic acids of the TCA cycle are decreased in leaf and fruit and some changes in metabolite and transcript profiles are similar to those observed in illuminated leaves of tomato deficient in TCA cycle enzyme expression Urbanczyk-Wochniak et al., 2006). Furthermore, the results presented here are comparable to those observed in the CMSII mutant of tobacco, which is deficient in complex I function, has a severely inhibited growth phenotype, and displayed constitutively high alternative oxidase activity associated in particular with high alternative NAD(P)H dehydrogenase activity (Dutilleul et al., 2003). However, some divergences exist since the CMSII mutant was able to permanently adjust the cell redox homeostasis, which was not the case in the P 35S :Slgalldh RNAi lines, at least in the case of ascorbate. Interestingly, our results corroborate the work of Millar et al. (2003) who showed that respiration controls ascorbate synthesis in plants, particularly through a link between complex I and GalLDH protein. In the transgenic tomato plants described here, the reduction in SlGalLDH activity appears to directly affect the efficiency of the alternative respiratory pathway and consequently the mitochondrial function, with the combination of changes in ascorbate and respiration resulting in a consequent alteration of the cell growth processes. Since several enzymes of the Kreb's cycle are redox regulated (Balmer et al., 2004), it is conceivable that any variation in the redox state of the plant may have additional effects on plant growth by modulating energy pathways.

Plant Material and Growth Conditions
Cherry tomato (Solanum lycopersicum 'West Virginia 106') plants were grown in a greenhouse with supplemental lighting when needed with a light period of 14/10 h, under a thermoperiod of 25°C/20°C, and watered daily three times with a solution at pH 5.8 containing oligoelements plus 3.5 mM KNO 3 , 1 mM K 2 SO 4 , 2 mM KH 2 PO 4 , 6 mM Ca(NO 3 ) 2 , and 2 mM MgSO 4 , until the stage fruit set of the first truss, and then 4 mM KNO 3 , 1.5 mM K 2 SO 4 , 1.5 mM KH 2 PO 4 , 4 mM Ca(NO 3 ) 2 , and 1.5 mM MgSO 4 . For in vitro culture, plants were grown on one-fourth Murashige and Skoog basal Murashige and Skoog medium (Kalys-Duchefa) containing 3% Suc under 14/10 h photoperiod at 400 mmol m 22 s 21 and at 25°C. Tomato fruits were harvested at various developmental stages according to the number of DPA and fruit diameter. Prior to all biochemical and molecular analysis, samples were quickly frozen in liquid nitrogen, ground to fine powders, and stored at 280°C until use.

Determination of Photosynthetic Activity
Photosynthetic activity of attached leaves was measured with a CO 2 analyzer using infrared detection (LCA3; Analytical Development Corporation).

Respiration Determinations of Isolated Mitochondria
To measure mitochondrial respiratory parameters, mitochondria were prepared from young leaf (50-100 g) following the method of Holtzapffel et al. (2002). After isolation and purifications steps, the mitochondria were washed and resuspended in the incubation medium containing 10 mM TES, pH 7.2, 300 mM Suc, 0.1% (w/v) bovine serum albumin (BSA), 5 mM Kh 2 PO 4 , 5 mM MgCl 2 , and 10 mM KCl to a concentration of 10 to 50 mg protein mL 21 . Respiration measurement was performed using O 2 electrode (782 Oxygen meter, Strathkelvin Instruments) in the presence of several electron donors and respiratory inhibitors like KCN and SHAM. Mitochondrial integrity was determined according to Holtzapffel et al. (2002) and was generally between 75% to 85%.

Cytological Analyses
Pericarp cell number and size were measured in fruits as described by Cheniclet et al. (2005). The cell size of leaflet epidermis was measured by applying a thin layer of collodion 4% resin (Merck) of about 2 to 3 cm 2 on the adaxial surface of the fourth and fifth leaves on the plant. After evaporation of the solvent, the epidermis imprint was visualized under a Zeiss Axiophot microscope with a Spot digital camera (Diagnostic Instruments). Cell size of leaflet epidermis was determined using the Image Pro-Plus software (Media Cybernetics). Leaflet surface was determined by scanning the same leaves used for epidermis imprint and the resulting image was analyzed using the Image Pro-Plus software.

RNA Extraction, Reverse Transcription-PCR Analysis, and DNA Construct Preparation
Total RNA extraction from tomato fruit and semiquantitative reverse transcription (RT)-PCR analysis using SlGalLDH gene-specific primers (Supplemental Table S1) were performed as described in Lemaire-Chamley et al. (2005). To obtain an SlGalLDH-specific cDNA fragment, RT (SuperScript II Reverse transcriptase) was performed using 2 mg of total RNA (Invitrogen, life technologies). A 133 bp SlGalLDH DNA fragment located in the 3#-untranslated region of the cDNA (GenBank accession no. AB080690) was then amplified by PCR using Ex Taq DNA polymerase (TaKaRa Bio INC) and the gene-specific primers GLDF4 and GLDR4. The purified DNA fragment was cloned as an inverted repeat under the control of the 35S promoter using the Gateway cloning system as described by Karimi et al. (2002), first into the entry vector (pDONR 201), then into the destination vector [pK7GWIWG2 (1)]. This construct was checked by sequencing and introduced into cherry tomato ('West Virginia 106') plants by Agrobacterium tumefaciens strain GV3101 according to Hamza and Chupeau (1993). The first generation plants (T0) were characterized and screened. Of the 12 lines studied, four lines named 2, 5, 8, and 38 were selected for further detailed analyses in the T1 generation. Control plants correspond to plants transformed with an empty vector following the same protocol.

Experimental Design for Transcriptome and Metabolome Analyses
Plant materials used for these studies were the leaflet of the fourth leaf and the outer pericarp of the orange fruit. One-centimeter-length young leaves were tagged at the plant apex to allow comparison of leaflets at the same age and position on the plant. For fruit studies, flowers were tagged at anthesis (fully opened flower) and fruit were further selected according to their color measured on the external surface of fruit using a chromameter. The Lab values corresponding to orange stage were, respectively, L45, a 5 approximately 20, b 5 approximately 30. Once harvested, the leaflets and the outer pericarp were combined in pools of 10 to 15 samples, frozen in N 2 , and stored at 280°C.
For transcript expression profiling of P 35S :Slgalldh RNAi silenced lines versus control plants, microarray experiments were performed using two biological repeats for each tissue (four slides per experiment) and technical duplicates with the dyes reversed (dye swap) as described in Lemaire-Chamley et al. (2005).

Hybridization and Data Acquisition
The TOM1 cDNA microarrays (Center for Gene Expression at the Boyce Thompson Institute; http://bti.cornell.edu/CGEP/CGEP.html) contain 13,400 printed elements corresponding to approximately 8,700 unigenes (Alba et al., 2004). The Cyscribe Post Labeling kit from Amersham (catalog no. RPN5660X, Amersham Biosciences) was used according to the manufacturer's recommendations, with 30 mg of total RNA per condition. The pooled cy3-and cy5-labeled cDNAs were then concentrated on Microcon YM-30 columns (Amicon Bioseparations, Millipore) and mixed with 90 mL of hybridization solution containing 1:1 (v:v) formamide (53 SSC, 0.25% SDS, 53 Denhardt's solution, and 1 mg mL 21 denatured salmon sperm DNA [Stratagene]). Slides were hybridized in an automatic hybridization station HS 4800 (Tecan) with a washing prerun in 13 SSC, 0.1% SDS, for 1 min. The probe solution was boiled for 1 min and then injected into the hybridization chamber. Slides were incubated at 42°C for 16 h, with medium agitation, and then washed sequentially at 30°C in 13 SSC, 0.1% SDS for 1 min, this step was repeated three times, in 0.13 SSC, 0.1% SDS for 1 min, three times, and finally in 0.13 SSC for 30 s. Slides were dried in the hybridization station for 3 min, with 2.7 bars of nitrogen gas. Microarray slides were scanned with a Genepix 4000 B fluorescence reader (Axon Instruments) using Genepix 4.0 image acquisition software with photomultiplier tube voltage adjusted to 500 V for Cy3 and 600 V for Cy5.

Bioinformatic Analysis
Spot flagging was done first by Genepix (missing spots) and then by visual inspection of the images to exclude the bad spots (saturation and heterogeneity). Raw data files were submitted to LIMMA v2.3.3 Bioconductor package (Smyth et al., 2005; http://bioinf.wehi.edu.au/limma/). Negatively flagged spots were excluded from further analysis by giving them a zero weight value. Data were then normalized by the print-tip lowess method without background subtraction followed by the scale method to adjust the data between the slides. The correlation between the replicated spots was calculated (duplicateCorrelation function) and the linear model was fitted with the lmFit function for each gene, using this correlation. Moderated t statistics and log odds of differential expression were computed (eBayes function) for the contrast of interest (P 35S :Slgalldh RNAi transgenic line versus wild type) and for each gene. The calculated P values for this contrast were adjusted for multiple testing with the false discovery rate method. All analyses were performed using the default parameter setting of LIMMA.

In Situ Hybridization
To use SlGalLDH as a probe, riboprobes were synthesized from plasmids containing a 506 bp fragment of SlGalLDH cDNA obtained from tomato by PCR using GLDF1 and GLDR1 primers (Supplemental Table S1). The sense and antisense digoxygenin-labeled riboprobes were generated by run off transcription using T7 and SP6 RNA polymerases according to the manufacturer's protocol (Roche Diagnostics). For in situ hybridization, tomato flower buds, fruits of 7 and 20 DPA, young leaves, and shoot and root tips were sampled and processed as described by Bereterbide et al. (2002).

Production of a Rabbit Polyclonal Antibody Anti-SlGalLDH and IgG Purification
For polyclonal antibody preparation, two synthetic polypeptides corresponding to the SlGalLDH protein were prepared by the Eurogentec Company. Their sequences that were deduced from the SlGalLDH tomato gene (accession no. AB080690) were: H2N-MSKEKGPPKNKPKYT1C-CONH2 (amino acids 321-335) and H2N-AYNQARKELDPNRIL1C-CONH2 (amino acids 559-573). The 1C corresponds to a Cys residue needed for the coupling with the keyhole limpet hemocyanin protein carrier. The antibodies were produced by rabbit immunized with a mix of the two peptides. Total IgGs were purified on affinity column (AF-amino TOYOPEARL 650 M) by Eurogentec.
Enzyme Activity, SDS-PAGE, and Immunodetection of SlGalLDH Fresh young leaves (approximately 0.5 g) were homogenized with a mortar and pestle in 1 mL of 0.1 M Tris-HCl buffer (pH 7.5) that contained 0.3 M Suc, 1% (w/v) polyvinylpolypyrrolidone, 0.2% (w/v) BSA, 1 mM dithiothreitol, and 50 mM EDTA. SlGalLDH activity was assayed spectrophotometrically at room temperature by measuring the increase in A 550 accompanied by the reduction of cytochrome c as described by Ô ba et al. (1995). Protein was quantified using BSA as a standard. Protein samples (10 mg per lane) were separated on 15% SDS-PAGE and transferred to Immobilon-P membrane prior to immunological detection by western-blot analysis using the SlGalLDH antibody at 1:300 dilution. Antigen-antibody complexes were detected with horseradish peroxidase-conjugated IgG diluted 1:10,000 (Chemicon) with the BM Chemiluminescence blotting substrate (POD) system from Roche Applied Science.

Ascorbic Acid Analysis
Ascorbate analyses were performed according to Leipner et al. (1997). Between 0.5 and 1 g of frozen samples were homogenized in 2.5 mL of cold 3% (w/v) metaphosphoric acid and 2.5 mM EDTA. The homogenate was then centrifuged at 10,000g for 10 min at 4°C. An aliquot of 200 mL was incubated for 15 min at room temperature with 100 mL of K 2 HPO 4 (45%), and either 50 mL of distilled water to measure the reduced ascorbate, or 50 mL of homo-Cys (0.1%) to measure the total ascorbate. After the incubation, 500 mL of citrate-P buffer (2 M, pH 2.3) were added. The A 524 was measured immediately after addition of 500 mL 2,6-dichloroindophenol (0.008% w/v).
For measurement of apoplastic ascorbate content, the intercellular washing fluid (IWF) was prepared using a method similar to that described by Turcsányi et al. (2000). Leaflets of the fourth leaf (about 1 g) were washed in distilled water and were twice vacuum infiltrated (270 kPa) for 3 min with 50 mL of 10 mM citrate buffer (pH 3.0) containing 100 mM KCl to reach maximum infiltration of cellular air spaces. Leaflets were then carefully blotted dry, rolled, and inserted into a 5 mL tip placed over a preweighed centrifuge tube containing 50 mL of 5% (w/v) cold metaphosphoric acid. The IWF (about 100 mL g 21 FW) was subsequently collected by centrifugation at 1,200g for 10 min at 4°C and immediately used for ascorbate measurement. The time from the harvest of the leaf to the beginning of centrifugation was less than 10 min. Activity of Glc-6-P dehydrogenase (EC 1.1.1.49), an enzyme located in the cytoplasm and the chloroplast stroma, was absent in the IWF, indicating that IWF was not contaminated by intracellular sap. To measure de novo ascorbate biosynthesis in leaves, 10 to 20 g of leaflet were rapidly washed in distilled water, sliced in fine stripes, and incubated at 25°C under continuous light (200 mmol photons m 22 s 21 ) in Murashige and Skoog medium buffered with 100 mM Tris-HCl (pH 8.0) in the presence of 25 mM gulono-1,4-lactone or galactono-1,4-lactone (approximately 200 mg tissue in 5 mL medium). In time course of the incubation, the tissues were wringed on paper and snap frozen in liquid N 2 , and stored at 280°C prior to ascorbate assay.

Determination of Levels of Other Metabolites
Samples of leaflets and orange fruit pericarp prepared as described above were used for metabolite extraction as described by Nunes-Nesi et al. (2005). The level of all metabolites was quantified by gas chromatography-mass spectrometry exactly following the protocol described by Roessner et al. (2001), with the exception that the peak identification was optimized to tomato tissues (Roessner-Tunali et al., 2003).
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number AB080690.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Table S1. Sets of PCR primers used to amplify specific regions of genes of interest.
Supplemental Table S2. Central metabolism of fully expanded leaf and orange fruit of P 35S :Slgalldh RNAi transgenic and control plants.