Altered Levels of Aroma and volatiles by Metabolic Engineering of Shikimate Pathway Genes in Tomato Fruits

The tomato (Solanum lycopersicum) fruit is an excellent source of antioxidants, dietary fibers, minerals and vitamins and therefore has been referred to as a “functional food”. Ripe tomato fruits produce a large number of specialized metabolites including volatile organic compounds. These volatiles serve as key components of the tomato fruit flavor, participate in plant pathogen and herbivore defense, and are used to attract seed dispersers. A major class of specialized metabolites is derived from the shikimate pathway followed by aromatic amino acid biosynthesis of phenylalanine, tyrosine and tryptophan. We attempted to modify tomato fruit flavor by overexpressing key regulatory genes in the shikimate pathway. Bacterial genes encoding feedback-insensitive variants of 3-Deoxy-D-Arabino-Heptulosonate 7-Phosphate Synthase (DAHPS; AroG209-9) and bi-functional Chorismate Mutase/Prephenate Dehydratase (CM/PDT; PheA12) were expressed under the control of a fruit-specific promoter. We crossed these transgenes to generate tomato plants expressing both the AroG209 and PheA12 genes. Overexpression of the AroG209-9 gene had a dramatic effect on the overall metabolic profile of the fruit, including enhanced levels of multiple volatile and non-volatile metabolites. In contrast, the PheA12 overexpression line exhibited minor metabolic effects compared to the wild type fruit. Co-expression of both the AroG209-9 and PheA12 genes in tomato resulted overall in a similar metabolic effect to that of expressing only the AroG209-9 gene. However, the aroma ranking attributes of the tomato fruits from PheA12//AroG209-9 were unique and different from those of the lines expressing a single gene, suggesting a contribution of the PheA12 gene to the overall metabolic profile. We suggest that expression of bacterial genes encoding


Introduction
Tomato fruits are an important source of vitamins, dietary fibers, minerals, and antioxidants in the human diet [1].During the ripening stage, tomatoes produce a large number of specialized metabolites, including volatile organic compounds [2,3], which serve as key components of the tomato fruit flavor [4,5].These compounds are utilized to attract seed dispersers [6], as part of the plant defense mechanisms against herbivores [7] and in plant-plant communication.They are derived from a diverse set of precursors including amino acids, fatty acids and carotenoids [4].The taste of a tomato is a result of the interactions of sugars, acids and a set of 20-30 volatile compounds.Among the several hundred volatile compounds accumulating in ripe tomato fruits, almost all of those related to flavor are derived from the essential amino acids phenylalanine (Phe), leucine (Leu) or isoleucine (Ile) [8].It has also been proposed that volatile compounds produced in the ripe tomato fruits act as sensory cues for nutritional and health values [4,6].
One of the major biosynthetic pathways in plants is the shikimate pathway which leads to the synthesis of three aromatic amino acids (AAA) Phe, tyrosine (Tyr) and tryptophan (Trp) [9].The shikimate pathway is a metabolic bridge between central carbon metabolism and specialized metabolism with regard to the regulation of AAA biosynthesis (Figure 1; [10,11]).In the last decade, many researches focused on exploring the enzymatic steps of the shikimate pathway, as well as the regulation of shikimate pathway enzymes.However, the carbon allocation towards the complex network of specialized metabolites that are derived from the AAA is not well understood [12,13].One of the major regulatory mechanisms of flux through metabolic pathways is enzyme feedback inhibition loops, in which the end product metabolite of a given metabolic pathway feedback inhibits the activity of one of the enzymes of this pathway.Such enzyme feedback inhibition loops are common in metabolic pathways of amino acids [14], and are used to balance the metabolic homeostasis of the cell, to reduce toxicity and to promote plant development [15,16,17].Plants containing modified, feedback-insensitive enzymes have been used to study the operation of metabolic pathways under uncontrolled conditions, and to produce plants tolerant to toxic combinations of amino acids or amino acid analogues, such as glyphosate (commercially known as Roundup; [18]).This also led to the identification of mutations in specific enzymes rendering them insensitive to feedback inhibition [14,17].
Overexpression of these feedback-insensitive genes allowed the plants to overcome the feedback inhibition of the native plant genes, "open" fluxes and change the carbon flow toward production of specialized metabolites.One such enzyme that was used encodes the first key enzyme of the shikimate pathway, namely, 3-Deoxy-D-Arabino-Heptulosonate 7-Phosphate Synthase (DAHPS; E. coli AroG) [19].Another gene encodes a bi-functional Chorismate Mutase/Prephenate Dehydratase (CM/PDT; E. coli PheA), that converts chorismate via prephenate into phenylpyruvate (Figure 1, [20]).Previously, we overexpressed a bacterial gene encoding a mutant feedback-insensitive DHAPS (AroG 209 ) in Arabidopsis thaliana plants as well as in tomato fruits.In general, overexpression of the bacterial feedback-insensitive AroG increased the synthesis of all AAA, causing considerable enhancements in the levels of multiple specialized metabolites in both Arabidopsis and tomato [19,21].In Arabidopsis vegetative tissues, constitutive expression of the bacterial AroG 209 caused massive metabolic changes and enhanced levels of phenylpropanoids, glucosinolates and phytohormones [19].In tomato fruits, on the other hand, expression of AroG 209 under a fruit specific E8 promoter led to enhanced levels of multiple volatile and nonvolatile compounds derived from phenylpropanoids, in addition to volatiles and carotenoids derived from terpenoids.These results demonstrate the complex network between those metabolic pathways [21].Furthermore, expression of the bacterial feedback-insensitive AroG in Petunia × hybrida cv "Mitchell Diploid" as well as expression of the endogenous PhDAHP1, resulted in dramatically increased levels of Phe and fragrant benzenoid-phenylpropanoid volatiles [12,22].
Phe is synthesized predominantly via the arogenate pathway [23], but the participation of the phenylpyruvate route in Phe biosynthesis has been previously suggested in Arabidopsis [24] and petunia [25].The PheA gene converts chorismate via prephentate into phenylpyruvate.Previously, expression of a gene encoding a mutated E.coli feedback-insensitive PheA in Arabidopsis plants was shown to impact the synthesis of AAA and the specialized metabolites derived from them [24].The PheA-expression lines had increased levels of specialized metabolites derived from Phe and Tyr, but reduced levels of specialized metabolites derived from Trp.These results imply regulatory cross-interactions between the flux of AAA biosynthesis from chorismate and their metabolism into various Phe specialized metabolites [24].However, the possibility of using feedback-insensitive PheA in fragrant rich tissues such as tomato fruits or petunia petals has yet to be investigated.The significant variation between the specialized metabolites accumulating in Arabidopsis suggested that both feedback-insensitive AroG and PheA can function as excellent metabolic engineering tools for the production of specialized metabolites.
We aimed to study the metabolic pathways involved in determining the tomato fruit flavor.The ultimate goal of our research was to identify genes that control the synthesis of the flavor volatiles and use this knowledge to produce a better-tasting tomato.We hypothesize that manipulating the DAHPS and CM/PDT key enzymes will have different effects on the carbon flux and will allow identification of metabolites that are usually below detection level.Our previous studies suggested that massive metabolic changes occur in the AroG 209-9 transgenic tomatoes.Here we first tested the metabolic changes of tomato fruits expressing the PheA gene or both the AroG 209-9 and PheA 12 genes (PheA 12 //AroG 209-9 ) crossed plants.We discovered that plants expressing both of the bacterial genes showed a unique metabolic profile that was predominantly impacted by expression of the AroG 209-9 gene, while the PheA 12 tomato line showed only minor metabolic differences.Our results suggest that this approach is suitable for altering tomato volatiles and aroma in a combinatorial manner.Furthermore, this method can be used for the discovery of new metabolic networks by increasing the abundance of compounds that are usually below detection levels in the wild type fruit.

Plant material and growth condition
Flowers of greenhouse-grown tomato plants were marked at anthesis, and ripe red fruits were harvested approximately 48 days post anthesis.Each biological repeat was a mixture of 3-5 individual fruits.After harvesting, the peel and flesh (without the gel and seeds) were manually dissected and frozen in liquid nitrogen [26].Homozygous plants of AroG 209-9 and PheA 12 were crossed and self-pollinated to generate the double transgenic cross line PheA 12 //AroG 209-9 (F1 generation).

Plasmid construction and tomato stable transformation.
Genomic DNA was isolated from Escherichia coli (K-12 strain) and two genes, the 3-Deoxy-D-arabino-Heptulosonate 7-Phosphate Synthase isoform G (DAHPS; AroG) and the Chorismate Mutase/Prephenate Dehydratase (CM/PDT; PheA) were amplified and cloned as previously described [19,21,24].The AroG and PheA genes were each fused to the tomato fruit-specific promoter E8 [27], whose expression is spatially and temporally regulated in mature tomato fruit [28].The genes were fused at the 5'-end to a plastid targeting signal peptide originated from RUBISCO small subunit [29].Primers are listed in Supplementary Table S1.Each of the chimeric genes was introduced into Agrobacterium tumefacies strain EHA-105 and used for plant transformation.Wild type tomato plants (M82 cultivar) were inoculated by submersing cotyledons in the transformed A. tumefacies culture as previously described [30,31].Tomato transformation and genotyping were performed by Hazera Ltd (www.hazera.co.il/).

LC-MS Metabolomics analysis
Non-targeted metabolic analysis was performed using 500 mg of tomato peel and flesh, extracted in 80% methanol.Sample preparation and injection conditions were performed as previously described [32].The analysis of the raw LC-MS (UPLC-qTOF-MS) data was performed using the XCMS software from the Bioconductor package (v.2.1) for the R statistical language (v.2.6.1) that performs chromatogram alignment, mass signal detection and peak integration [33].XCMS was used with the following parameters: fwhm = 10.8, step = 0.05, steps = 4, mzdiff = 0.07, snthresh = 8, max = 1000.Injections of samples in the positive and negative ionization modes were performed in separate injection sets and pre-processing was done for each ionization mode independently.Differential mass ions were determined using a Student's t-test (JMP software) and 17 differential metabolites were subsequently assigned.Principal Component Analysis (PCA) plot and ANOVA tests were performed by the T-MEV4 software [34,35].A Student's t-test analysis was performed on metabolites level using the JMP software (SAS).Volume 2, Issue 2, 75-92.

Detection and profiling of volatile aroma compounds using GC-MS
GC-MS analysis of polar volatile compounds was carried out as previously described [21,36].Briefly, a mix of 2-5 fruits (10 g) were harvested at the ripening stage (flesh and peel) and extracted with 30 ml MTBE:hexane (1:1) containing 2 g isobutylbenzene as an internal standard.Following overnight incubation with shaking at 150 rpm, the extract was centrifuged at 10,500 g for 10 min and the supernatant was passed through a 0.2 μm filter.Samples were evaporated, using nitrogen, to a final volume of 200 μl before injection into a GC-MS instrument [19].Identification of the compounds was based on a comparison of mass spectra and retention times with those of authentic standards (Sigma, Milwaukee, WI, USA) analyzed under similar conditions [21,37].Statistical analysis was performed using the JMP software (SAS).

Sensory panel and evaluation of samples
A panel of 10 trained flavor specialists evaluated the aroma of samples by smelling the fruits as previously described [21].Preliminary tests were carried out to improve the ability of the assessors to recognize odor changes and consistently quantify sensory properties.The panelists had been trained in the quantitative description of tomato attributes according to selection trials based on French norms (ISO8586-1, AFNOR V09-003).For each fruit sample, cut sections containing all fruit tissues were used for aroma evaluation by the panel.Values of individual fruits were ranked from zero (none) to 5 (very strong).The professional attributes considered were the following: acidic, floral, fresh, green, metallic musty, ripe, spicy, sweet and overall aroma intensity.

Generating tomato fruits expressing feedback insensitive enzymes of the shikimate pathway
Our aim was to direct the central carbon metabolism through the shikimate pathway into specialized metabolites derived from the AAA.Therefore, our focus was on the first enzyme of the shikimate pathway, DAHPS, and the first and the second enzymes of Phe biosynthesis, CM/PDT.However, these enzymes are post translationally feedback regulated by their AAA products (Figure1).To overcome this regulation, we isolated, engineered and cloned two bacterial orthologues genes from Escherichia coli, and manipulated them to generate feedback insensitive isoforms of the genes.The E. coli, DAHPS gene, AroG, was point mutated in position 209.This mutation completely abolishes the Phe inhibition of the AroG allosteric site [38].In the second enzyme, the E. coli bi-functional CM/PDT, PheA, the catalytic activities of the CM and PDT domains are located at amino acids 1-300 while the C-terminal domain is responsible for the allosteric feedback inhibition by Phe [39].A truncated CM/PDT protein lacking the C terminus allosteric site retained the CM and PDT activities but did not exhibit feedback inhibition by Phe, and resulted in over accumulation of Phe [24,39].A chimeric gene encoding the bacterial PheA under a tomato E8 fruit specific promoter was constructed and transformed into tomato plants.After selection of kanamycin resistant plants, the expression of the PheA gene in tomato fruit was verified by semi-quantitative RT-PCR (Figure S1).For each transgene, five independent transformation events were collected and subjected to LC-MS analysis, using both the negative and positive ion modes.This analysis revealed 5,723 and 7,125 mass signals, respectively.As shown in the PCA plots in Supplementary Figure S1, the most separated PheA transformation event was PheA 12 in both the negative and positive ion modes (Figure S2A and S2B respectively).As shown in Table S2, Phe level was significantly induced by more than two fold in PheA 6 and PheA 12 while Tyr and Trp were not changed.In addition, we selected unknown mass signals that were significantly altered mainly in PheA 6 and PheA 12 lines.Based on these results, the PheA 12 line was chosen for further work.As for the AroG gene, point mutation isoform 209, was fused to the E8 promoter and transformed in M82 tomato and transformation event number 9 was selected as previously characterized [21].In order to detect the combined effect of the two shikimate pathway enzymes on tomato flavor, the AroG 209-9 and PheA 12 lines were crossed.AroG 209-9, PheA 12 and the cross PheA 12 //AroG 209-9 line were analyzed and compared to the wild type (WT) M82 control.

Effect of the AroG 209-9 and PheA 12 genes on the level of non-volatile metabolites in tomato fruit tissues
Fleshy tomato fruits develop in several marked stages, each possessing a characteristic metabolic profile [40].Dramatic metabolic alteration of central metabolites occurs during fruit maturation [21,41] followed by induction of fruit flavor [4].Therefore, we focused our study on ripe red tomato, which has the highest level of flavor volatiles.Fruits were separated into two tissues: (i) peel, which is typically composed of multiple cell types, including epidermis, collenchyma, and some parenchyma, and (ii) flesh, which refers to the pericarp material from which the peel has been removed and therefore is predominantly composed of parenchyma and collenchyma [32].The transgenic tomato plants did not exhibit any morphological differences, and their course of development, including color breaking and ripening time was similar to that of the control plants (data not shown).Ripe fruits were harvested from AroG 209-9 , PheA 12 , PheA 12 //AroG 209-9 and WT plants and analyzed by an established high-resolution LC-MS-based metabolomics platform (negative ion-mode; [42]).To get a global view on the metabolic effects, the LC-MS mass signals dataset was analyzed using PCA plots.As shown in Figure 2  To identify compounds whose levels differed between the transgenic lines, we compared the LC-MS mass signals to known chemical standards and libraries.As summarized in Table 1, 17 compounds were identified in the flesh and peel of the tomato fruits.In the flesh tissue (Table 1A), the levels of Phe and Tyr were increased in both AroG 209-9 and PheA 12 //AroG 209-9 compared to the WT fruits.Trp level was increased only in the AroG 209-9 but not in the PheA 12 //AroG 209-9 line.In addition, the level of isopropylmalic acid was increased in the flesh of both the AroG 209-9 and PheA 12 //AroG 209-9 lines.The levels of caffeic acid-hexose and dicaffeoylquinic acid were significantly decreased in the flesh of both AroG 209-9 and PheA 12 //AroG 209-9 transgene fruits.The levels of naringenin chalcone-hexose and phloretin-di-C-hexose were significantly decreased in the AroG 209-9 while the levels of hydroxylated naringenin-hexose and naringenin hexose were significantly increased in PheA 12 //AroG 209-9 only.None of the identified compounds were significantly changed in the flesh of the PheA 12 line relative to the WT.

Table 1. List of the metabolites identified by LC-MS in the peel (A) and flesh (B) of the
In the peel tissue (Table 1B), the levels of all AAA, but most noticeably Phe and Tyr were increased in both the AroG 209-9 and the PheA 12 //AroG 209-9 lines, as well as isopropylmalic acid, coumaric acid-hexose-I, and tricaffeoylquininc acid.The levels of caffeic acid-hexose, ferulic acid hexose and phloretin-di hexose were significantly decreased in both AroG 209-9 and PheA 12 //AroG 209-9 lines.The levels of 3-caffeoylquinic acid, dicaffeoylquinic acid and naringenin chalcone-hexose were altered in the AroG 209-9 only.The level of naringenin hexose highly increased while coumaric acid hexose II deceased in PheA 12 //AroG 209-9 only.Similarly to the flesh tissues (Table A1), none of the identified compounds were significantly changed in PheA 12 line.Interestingly, the induction of AAA in both flesh and peel was higher in the AroG 209-9 line compared to the PheA 12 //AroG 209-9 line.In contrast, naringenin hexose levels were higher in the PheA 12 //AroG 209-9 than in the AroG 209-9 line.These findings suggest that although no major changes were found in the level of the identified compounds in the PheA 12 line, its combined expression with AroG 209-9 in the PheA 12 //AroG 209-9 line does affect the metabolic flux.

Effects of the expression of the AroG 209-9 , the PheA 12 gene or both genes on volatile metabolites and fruit aroma
Our aim was to detect the changes in the level of volatile compounds in ripe red fruits expressing the AroG 209-9, the PheA 12 gene or both genes.To that end, volatile metabolites accumulating in the tissue were extracted and analyzed by GC-MS as presented in Figure 3.The volatile compounds benzaldehyde, phenylacetaldehyde and phenylacetic acid were significantly increased in the AroG 209-9 and PheA 12 //AroG 209-9 lines, compared to PheA 12 and WT.However, the level of phenylacetaldehyde was higher in the PheA 12 //AroG 209-9 line than in the AroG 209-9 line.The levels of five other volatiles, eugenol (derived from p-coumaric acid; Figure 1), geranylacetone, β-ionone and limonene (derived from terpenoid) as well as benzaldehyde-4-methyl were reduced in the AroG 209-9 line, compared to the PheA 12 line and the WT.The PheA 12 //AroG 209-9 line showed similar reduction of eugenol, geranylacetone, benzaldehyde-4-methyl and limonene, (derived from terpenoid) but no significant reduction in β-ionone and limonene compared to WT.Except for a slight increase in the eugenol and limonene levels, the PheA 12 tomato line showed no differences in the identified volatile compounds relative to the WT tomato fruits.Tomato aroma was also evaluated by a sensory panel.This test provides human insight into the flavor experience, where sensory attributes, preferences, and decisions can be statistically related to the chemical components in foods [43].The organoleptic test was performed by a group of professionals trained in the quantitative description of tomatoes.Whole red fruits from each line were cut in half and evaluated by the various panel members by sniffing the samples (Figure 4).As shown in Figure 4, the panel suggested that the "acidic" aroma attribute was increased in the PheA 12 //AroG 209-9 tomato fruits.The floral aroma attribute was increased only in the AroG 209-9 tomato as previously described [21], but not in the PheA 12 or PheA 12 //AroG 209-9 fruits.In addition, it was suggested that the global aroma intensity of both PheA 12 and PheA 12 //AroG 209-9 tomato fruits was decreased as compared to the AroG 209-9 tomato but not in comparison to the WT fruits.The other attributes were similar between the lines.

Discussion
The results presented in this study suggest that overexpression of feedback insensitive mutants of AroG 209-9 //PheA 12 genes generated new tomato aroma composition (Figure 4).Previous research used ectopic expression of the petunia MYB transcription factor ODORANT1 to alter the expression of a set of genes related to Phe metabolism in tomato fruits including DAHPS, CM and Arogenate/Prephenate Dehydratase [44].This increased metabolic activity, was coupled to a considerably enhanced flux through part of the phenylpropanoid pathway, but did not result in induction of Phe-derived flavor volatiles, suggesting that factors beyond substrate availability limit their synthesis such as post-transcriptional regulation [44].We chose a different approach using bacterial feedback-insensitive variants of biosynthetic enzymes in order to bypass enzyme feedback inhibition loops of the shikimate and AAA biosynthesis pathways (Figure 1).In general, the expression of these genes resulted in alteration in the production of several volatile as well as non-volatile compounds and in changes in fruit aroma.This emphasizes the importance of using ectopic expression of feedback-insensitive enzymes for metabolic engineering.
PheA 12 , which is a feedback-insensitive isoform of the first and second enzymatic steps of the Phe biosynthesis, caused only minor metabolic changes in tomato (Figure 2-4 and Table 1).Although PCA analysis showed separation between PheA 12 and WT samples in either flesh or peel tissue (Figure 2), none of the identified compounds were differentially changed in this line (Table 1).This may be due to the small subset of metabolites that we were able to identify (17 non volatiles and 9 volatiles, see Table 1 and Figure 3, respectively).Volatile analysis suggested slight induction of eugenol and limonene compared to control plants, although the induction was not significant (Figure 4).Taken together, these results suggest that feedback-insensitive PheA does not play a major role in AAA metabolism and specialized metabolite derived from it in tomato fruits.
On the other hand, AroG 209-9 , which is a feedback-insensitive form of the first enzymatic step of the shikimate pathway, caused massive metabolic changes in tomato (Figure 2-4 and Table 1).Similar results were previously shown in Arabidopsis vegetative tissues and petunia petals [12,21].In order to enhance the carbon flux of AAA biosynthesis and direct it toward Phe derived volatiles, we crossed PheA 12 and AroG 209 tomato plants.The PCA plot separation showed close clustering of AroG 209-9 with the PheA 12 //AroG 209-9 samples of peel and flesh tissue however, the PCA samples did not overlap in either tissue (Figure 2).Interestingly, several compounds had different levels in the AroG 209-9 fruits compared to the crossed tomato line (Figure 3 and Table 1).This implies that tomato fruits expressing both the PheA 12 and AroG 209-9 genes have a metabolic profile that is predominantly, but not solely, impacted by the expression of the AroG 209-9 gene.
Because volatiles are derived from a diverse set of precursors, including amino acids, fatty acids and carotenoids, changes in the level of these different precursors can impact tomato aroma.Phe and Tyr were significantly increased in the flesh of both the AroG 209-9 and PheA 12 //AroG 209-9 lines, however the increase in their levels was higher in the AroG 209-9 line than in the cross line: Phe 7.68 and 5.39 and Tyr 23.2 and 9.34, fold changes respectively (Table 1A).This effect was also seen in the levels of AAA in the peel (Phe 15.94 and 12.22, Tyr 51.52 and 21.51 and Trp 2.37 and 1.68, fold changes respectively (Table 1B).The overall changes in Phe and Tyr are much more pronounced than in Trp in the peel and flesh of both the AroG 209-9 and the PheA 12 //AroG 209-9 lines.As the levels of Phe and Tyr were higher in the AroG 209-9 line (Table 1), the effect on the aroma profile of the fruit might be different in these two lines, mainly due to Phe-derived volatiles (Figure 3-4).A Phe derived volatile compound, phenylacetaldehyde, was higher in the cross line than in the AroG 209-9 line (Figure 3).This may indicate additional regulatory metabolic steps of the Phe derived metabolites.In petunia, it's been demonstrated that phenylacetaldehyde is synthesized via Phe and phenylpyruvate [45].Therefore, we suggest that in tomatoes the high induction of phenylacetaldehyde levels in the PheA 12 //AroG 209-9 line might be due PheA contribution that drives Phe synthesis via phenylpyruvate (Figure 1).Carotenoids-derived precursor's β-ionone and geranylacetone [4] were reduced in both AroG 209-9 and PheA 12 //AroG 209-9 lines (Figure 3) supporting cross-talk between phenylpropanoid and carotenoid (terpenoid) derived volatiles [46].
While several hundred volatiles have been identified in tomato, only a set of 20-30 volatile compounds actively contribute to tomato flavor; they are present in sufficient quantities to noticeably stimulate the olfactory system [4].This threshold is determined by both the concentration of the substance and the organism's ability to detect it [47].Flavor thresholds vary markedly between individuals and can be greatly influenced by the way in which the volatile is presented.The differences in volatile compounds between the lines (Figure 3) can give a varied tomato flavor (Figure 4).

Conclusion
This work demonstrates the potential for using feedback-insensitive enzymes to modify fruit flavor.Expression of feedback-insensitive enzymes in a combinatorial manner generates new aroma and affect flavor attributes.This approach is suitable for the enhancement of tomato volatiles and aromas as well as for the discovery of new metabolites and their biosynthesis and can be used in other plant species.

Figure 1 .
Figure 1.Scheme of the shikimate and AAA biosynthesis pathways in plants.Dashed gray lines represent post-transcriptional regulation of the DAHPS (bacterial AroG) and CM/PDT (bacterial PheA) by the AAA.Dashed black lines represent multiple enzymatic steps.Solid gray line represents biosynthetic pathway suggested by Kaminaga et al. 2006.
, the samples clustered into four groups: (i) flesh samples of the AroG 209-9 and PheA 12 //AroG 209-9 lines; (ii) flesh samples of the PheA 12 and WT lines; (iii) peel samples of the AroG 209-9 and PheA 12 //AroG 209-9 lines and, (iv) peel samples of the PheA 12 and WT lines.The AroG 209-9 and PheA 12 //AroG 209-9 samples were closer to each other than to WT and PheA 12 but still did not overlap.This suggested that the major metabolic effects in the transgenic fruits are due to the presence or absence of the AroG 209 gene, while the PheA 12 gene has a minor contribution to the overall metabolic profile.

Figure 3 .
Figure 3. Internal pools of volatile compounds altered in red tomato fruits expressing the AroG 209-9 (AroG), PheA 12 (PheA) or both genes PheA 12 //AroG 209-9 (Cross) line.Fruit samples were analyzed by GC-MS.Bars on top of the histograms indicate standard errors (n = 4).Letters indicate statistically significant differences between the three lines and WT using ANOVA all pairs comparison Tukey-Kramer HSD test.The AroG 209-9 data was previously published in Tzin et al. 2013.

Figure 4 .
Figure 4.A sensory profile of red ripe tomatoes expressing the AroG 209-9 (AroG), PheA 12 (PheA) and both PheA 12 //AroG 209-9 genes (Cross) line.Each attribute was scored on a scale between zero (none) to five (very strong).Bars on top of the histograms indicate standard errors (n = 4).Letters indicate statistically significant differences between the three lines and the WT using ANOVA all pairs comparison Tukey-Kramer HSD test.The AroG 209-9 data was previously published in Tzin et al. 2013.

Figure S2 :
Figure S2: PCA plot of the five transformation events of PheA (T1) homozygous lines of tomato.Mass signals of chemical compounds were obtained by UPLC-QTOF-MS in the negative and positive ion modes (5,723 and 7,125 mass signals, respectively) from the tomato red fruit (n = 4-7; T1 generation).The PCA plot was done on the Log10 values and generated using the T-MEV4 software.