Geranyl diphosphate (GPP), the precursor of most monoterpenes, is synthesized in plastids from dimethylallyl diphosphate and isopentenyl diphosphate by GPP synthases (GPPSs). In heterodimeric GPPSs, a non-catalytic small subunit (GPPS-SSU) interacts with a catalytic large subunit, such as geranylgeranyl diphosphate synthase, and determines its product specificity. Here, snapdragon (Antirrhinum majus) GPPS-SSU was over-expressed in tomato fruits under the control of the fruit ripening-specific polygalacturonase promoter to divert the metabolic flux from carotenoid formation towards GPP and monoterpene biosynthesis. Transgenic tomato fruits produced monoterpenes, including geraniol, geranial, neral, citronellol and citronellal, while exhibiting reduced carotenoid content. Co-expression of the Ocimum basilicum geraniol synthase (GES) gene with snapdragon GPPS-SSU led to a more than threefold increase in monoterpene formation in tomato fruits relative to the parental GES line, indicating that the produced GPP can be used by plastidic monoterpene synthases. Co-expression of snapdragon GPPS-SSU with the O. basilicum α–zingiberene synthase (ZIS) gene encoding a cytosolic terpene synthase that has been shown to possess both sesqui- and monoterpene synthase activities resulted in increased levels of ZIS-derived monoterpene products compared to fruits expressing ZIS alone. These results suggest that re-direction of the metabolic flux towards GPP in plastids also increases the cytosolic pool of GPP available for monoterpene synthesis in this compartment via GPP export from plastids.
Terpenoids represent one of the most diverse class of plant metabolites, being involved in numerous ubiquitous basic processes including photosynthesis, respiration, growth and development (Gershenzon and Kreis, 1999; Rodriguez-Concepción and Boronat, 2002). In addition to such vital molecules as sterols, carotenoids and the hormones gibberellins, strigolactones, abscisic acid and brassinosteroids, this class of metabolites includes monoterpenes, sesquiterpenes and diterpenes, which play important roles in plant defense against herbivores and pathogens, as well as in plant reproduction by attracting pollinators and seed dispersers (Dudareva et al., 2006). All terpenoids are derived from the universal five-carbon building blocks isopentenyl diphosphate (IPP) and its isomer dimethylallyl diphosphate (DMAPP), which are synthesized in plants by two alternative pathways that are localized in different subcellular compartments (Ashour et al., 2010; Hemmerlin et al., 2012). The classical mevalonic acid (MVA) pathway, which is localized in the cytosol and partially in peroxisomes, gives rise to IPP and via enzymatic isomerization to DMAPP, both of which serve as substrates (at a 2:1 ratio) for cytosolic farnesyl diphosphate (FPP) synthase (FPPS) to form FPP. In contrast, the plastidic methylerythritol phosphate (MEP) pathway directly produces both IPP and DMAPP (at a 6:1 ratio) for downstream formation of geranyl diphosphate (GPP) and geranylgeranyl diphosphate (GGPP) by the plastidic enzymes GPP synthase (GPPS) and GGPP synthase (GGPPS), respectively. While FPP in the cytosol serves as a precursor for sterols/brassinosteroids, plastidic GGPP is utilized for chlorophyll, carotenoid, strigolactone, abscisic acid and gibberellin biosynthesis. Independently, cytosolic sesquiterpene synthases use FPP and plastidic mono- and diterpene synthases use GPP and GGPP, respectively, as substrates. Although separated in two subcellular compartments, there is substantial evidence for metabolic interaction between the MVA and MEP pathways, with IPP exchange potentially in both directions (Kasahara et al., 2002; Nagata et al., 2002; Hemmerlin et al., 2003a; Laule et al., 2003; Schuhr et al., 2003; Dudareva et al., 2005; Furumoto et al., 2011). It was shown that the metabolic flux through the MEP pathway often exceeds that of the MVA pathway (Dudareva et al., 2005; Wu et al., 2006), and export of MEP pathway-derived IPP is of particular importance for cytosolic terpene biosynthesis in some plants (Adam et al., 1999; Steliopoulos et al., 2002; Dudareva et al., 2005; Hampel et al., 2005a,b; Orlova et al., 2009). Trafficking of IPP across the inner envelope membrane of plastids is mediated by an as yet unidentified metabolite transporter (Soler et al., 1993; Bick and Lange, 2003; Flügge and Gao, 2005).
While plastidic GGPPSs are homodimeric (Dogbo and Camara, 1987; Laferrière and Beyer, 1991; Burke and Croteau, 2002), GPPSs have either homodimeric or heterodimeric architectures depending on the plant species (Nagegowda, 2010). Heterodimeric GPPSs, like the one found in snapdragon (Antirrhinum majus), contain a catalytically inactive small subunit (GPPS-SSU) that interacts with a bona fide GGPPS subunit to form a heterodimer that catalyzes GPP formation (Tholl et al., 2004; Wang and Dixon, 2009). Moreover, inactive GPPS-SSUs interact in vitro and in planta with GGPPSs from distant plant species to form functional heterodimeric GPPSs (Burke and Croteau, 2002; Tholl et al., 2004; Orlova et al., 2009; Wang and Dixon, 2009). With the exception of Lithospermum erythrorhizon (Sommer et al., 1995; Li et al., 1998), all known plant GPPS enzymes appear to be plastidic. However, some cytosolic sesquiterpene synthases, in addition to FPP, can use GPP as a substrate to form monoterpenes (see the comprehensive list in Table S1). In cultivated strawberry (Fragaria ananassa) fruits, the cytosolic terpene synthase FaNES1 produces roughly similar amounts of the sesquiterpene nerolidol and the monoterpene linalool (Aharoni et al., 2004). Moreover, over-expression in the cytosol of monoterpene synthases (Ohara et al., 2003; Wu et al., 2006) and a sesquiterpene synthase possessing monoterpene synthase activity (Davidovich-Rikanati et al., 2008) resulted in formation of monoterpenes in transgenic plants, suggesting the existence of a GPP pool in this cellular compartment. These data raise questions about the origin of cytosolic GPP and its potential transport between plastids and the cytosol. To address these questions, a metabolic engineering approach was used to co-express snapdragon GPPS-SSU with the geraniol synthase gene GES, encoding a plastidic monoterpene synthase (Davidovich-Rikanati et al., 2007), and the zingiberene synthase gene ZIS, encoding a cytosolic terpene synthase that has been shown to possess both sesqui- and monoterpene synthase activities (Davidovich-Rikanati et al., 2008) in ripening tomato (Solanum lycopersicum) fruits under the control of the polygalacturonase (PG) promoter. A significant increase in monoterpenes was observed in both cases, suggesting that plastid-produced GPP is not only used for plastidic monoterpene formation, but is also exported to support cytosolic monoterpene biosynthesis.
Over-expression of snapdragon GPPS-SSU enhances monoterpene formation in tomato fruits
To increase the plastidic GPP pool available for monoterpene biosynthesis, we diverted the metabolic flux from carotenoid formation, which is highly active in ripening tomato fruits, by over-expressing the snapdragon GPPS-SSU gene under control of the fruit ripening-specific PG promoter (Nicholass et al., 1995). This promoter was chosen to limit GPPS-SSU expression to ripening fruits and avoid potential negative effects on vegetative tissues and general plant development (Orlova et al., 2009). Four independent lines with different GPPS-SSU expression levels, as determined by quantitative real-time PCR (Figure 1a) and protein gel-blot analysis (Figure 1b), were chosen for further characterization. GPPS-SSU transcript and protein levels were highest in line E9, intermediate in lines B3 and B10, low in line D8, and undetectable in the untransformed MP–1 control (Figure 1a,b). Consistent with the GPPS-SSU expression profiles, GPPS activity increased across transgenic fruits by 1.6- to 2.2-fold relative to controls containing endogenous GPPS activity (Table 1). To date, the exact architecture of tomato GPPS is still unknown; however, the tomato genome contains a gene encoding Sl-GPPS-SSU [Solanum lycopersicum (SI)], which belongs to a recently described SSU II sub-family (Wang and Dixon, 2009), and its expression did not change in transgenic fruits compared to control (Figure S1). An earlier proposed homodimeric tomato GPPS (van Schie et al., 2007a) is highly homologous to the Arabidopsis At2g34630 gene product, which was recently identified as a trans-type polyprenyl pyrophosphate synthase (PPPS; Hsieh et al., 2011; Ducluzeau et al., 2012) and is not expressed in tomato fruits based on quantitative real-time PCR using gene-specific primers.
Table 1. Prenyltransferase activities in crude extracts from control and GPPS-SSU tomato fruits (Br + 3)
Values are pkat mg−1 protein (means ± SE, n = 3–6).
Asterisks indicate statistically significant differences compared with control (MP–1): *P < 0.03, **P < 0.01 by Student's t test.
As expected from a PG promoter-driven expression, no GPPS-SSU transcripts were detected in fruits at the mature green (MG) stage (Figure S2a). GPPS-SSU transcripts began to accumulate at the breaker (Br) stage, reached the highest level 3 days after the Br stage (Br + 3), and declined in ripe fruits (Br + 10). GPPS-SSU over-expression reduced carotenoid accumulation, which was visible at various ripening stages (Figure S2b). Indeed, levels of trans-lycopene, the most abundant carotenoid in tomato fruits, decreased by 69–90% in transgenic fruits relative to control (Figure 1c and Figure S3b), showing negative correlation with GPPS-SSU expression. However, the levels of some lycopene-derived carotenoids, including β–carotene, zeaxanthin and lutein (Figure S3a), remained unaltered in the GPPS-SSU transgenic fruits (Figure S3b). The reduction in lycopene levels was not due to down-regulation of expression of the MEP pathway genes encoding 1–deoxy-d–xylulose 5–phosphate synthase (DXS) and 1–deoxy-d–xylulose 5–phosphate reductoisomerase (DXR), or those encoding the carotenoid biosynthetic enzymes phytoene synthase (PSY–1 and PSY–2) and phytoene desaturase (PDS), whose transcript levels were similar in transgenic and control fruits (Figure S1). Of two GGPPS genes in the tomato genome, only one, GGPPS–2, is expressed in fruits (Ament et al., 2006). Interestingly, its expression was up-regulated in the GPPS-SSU transgenic line (Figure S1), but there were no changes in GGPPS activity (Table 1).
To examine the effect of GPPS-SSU expression on the level of monoterpenes, volatiles were collected from ripe tomato fruits (Br + 10) for 24 h and analyzed by GC–MS. Marked differences in monoterpene profiles were observed between transgenic and control fruits (Figure 2a). Transgenic fruits produced geranial, neral and citronellol, and, to a lesser extent, geraniol, citronellal and linalool, but all these volatiles were absent in control fruits, except for a small amount of geranial (Figure 2a and Table S2). Internal pools of geraniol, citronellol, neral and geranic acid were also significantly increased in ripe GPPS-SSU transgenic tomato fruits, but the level of geranial increased only in the highly expressing E9 line (Figure 2b and Table S2). The amounts of monoterpenes, both those collected as emitted volatiles and directly extracted from fruits (Figure 2c,d and Table S2), positively correlated with the GPPS-SSU transcript and protein levels in the various transgenic lines (Figure 1a,b). These results indicate that the introduced snapdragon GPPS-SSU in tomato fruits diverted the metabolic flux of the plastidic terpenoid pathway from carotenoids towards GPP, which was subsequently converted to various monoterpenes.
Co-expression of snapdragon GPPS-SSU with lemon basil GES significantly increases monoterpene accumulation in tomato fruits
Recently, it was shown that the tomato genome contains 44 terpene synthase (TPS) genes, 26 of which are functional (Bleeker et al., 2011; Falara et al., 2011). Among the six TPSs found to be expressed in ripe tomato fruits, only one monoterpene synthase, TPS5/MTS1, uses GPP as a substrate to produce linalool (van Schie et al., 2007b). While a small amount of linalool was indeed found in GPPS-SSU-expressing tomato fruits (Table S2), the formation of other produced monoterpenes (Figure 2) cannot be attributed to TPS5/MTS1 action and is probably the result of the action of endogenous enzymes metabolizing GPP.
The low level of endogenous monoterpene synthases capable of utilizing GPP, and the presence of endogenous enzymes that use GPP in ripening tomato fruits, raise the question of whether the increased plastidic GPP pool in GPPS-SSU transgenic fruits is available for enzymatic conversion by introduced monoterpene synthases. Therefore, line E9 with the highest GPPS-SSU expression was crossed with a transgenic tomato line over-expressing O. basilicum GES under the control of the PG promoter (Davidovich-Rikanati et al., 2007). In the resulting F1 plants (GPPS-SSU × GES), the GPPS-SSU and GES transcript levels were comparable with those in fruits of parental lines (Figure 3a). Analysis of volatiles produced by ripe tomato fruits (Br + 10) of parental lines, as well as the cross, revealed the presence of the same monoterpene compounds, including citronellol, citronellal, geraniol, geranial and neral, (Figure 3b), with the exception of citronellyl acetate, which was not detected in GPPS-SSU fruits. While the composition of the monoterpene profile produced by fruits of the three analyzed lines was practically identical, the absolute amounts differed drastically between them, with the lowest amounts found with GPPS-SSU fruits and the highest amounts with GPPS-SSU × GES fruits (Figure 3b). Quantitative analysis of monoterpenes collected from GPPS-SSU × GES fruits (Figure 3c and Table S2) revealed on average 6.9- and 19.2-fold increases in geraniol, 2.8- and 7.5-fold increases in geranial, 3.0- and 8.6-fold increases in neral, 3.3- and 14.4-fold increases in citronellol and 3.5- and 5.4-fold increases in citronellal when compared to fruits of the GES and GPPS-SSU parental lines, respectively. Likewise, quantitative analysis of internal monoterpene pools accumulated in GPPS-SSU × GES fruits (Figure 3d and Table S2) revealed on average 7-fold increase in geraniol, geranic acid and citronellol, as well as 2.9- and 4.4-fold increases in geranial and neral, respectively, relative to fruits of the GPPS-SSU parental line. However, when GPPS-SSU × GES fruits were compared with the GES parental line, no statistically significant differences in internal monoterpene pools were detected (Figure 3d). The GES parental line also produced a small amount of myrcene, limonene and ocimene of unknown origin (Davidovich-Rikanati et al., 2007), the levels of which remained unchanged in GPPS-SSU × GES fruits (Table S2).
Co-expression of snapdragon GPPS-SSU with lemon basil ZIS increases cytosolic monoterpene formation in tomato fruits
It is well established that the MEP pathway provides precursors not only for plastidic monoterpene formation, but also supports terpene biosynthesis in the cytosol via export of IPP across the plastid envelope membranes (Dudareva et al., 2005; Hampel et al., 2005a,b; Orlova et al., 2009). As some cytosolic bifunctional terpene synthases exhibit both mono- and sesquiterpene synthase activities, we analyzed whether increased plastidic GPP formation also affects monoterpene biosynthesis in the cytosol. The GPPS-SSU line E9 was crossed with a transgenic tomato line over-expressing O. basilicum ZIS, encoding a cytosolic terpene synthase that not only uses FPP to form α–zingiberene and several other sesquiterpenes, but also accepts GPP as a substrate to produce a number of monoterpenes in vitro as well as in planta (Davidovich-Rikanati et al., 2008). In the resulting F1 plants (GPPS-SSU × ZIS), the GPPS-SSU transcript levels were not significantly different from those in the GPPS-SSU parental line (Figure 4a), while ZIS mRNA levels were reduced relative to the ZIS parental line (Figure 4a), probably due to the presence of several segregating copies of the ZIS transgene in the parental line. Analysis of volatiles collected from ripe tomato fruits (Br + 10 stage) of ZIS parental and GPPS-SSU × ZIS lines revealed that the sesquiterpene profiles were identical (Figure 4b), including 7–epi–sesquithujene, E–α–bergamotene, α–zingiberene and β–bisabolene. However, the amounts of sesquiterpenes emitted from GPPS-SSU × ZIS fruits, as well as the internal pools of sesquiterpenes, were reduced on average by 63–95% (Figure 4c and Table S3) and 42–83% (Figure 4d and Table S3), respectively, relative to the ZIS parental line, consistent with the 86% reduction in ZIS transcript levels in the GPPS-SSU × ZIS cross (Figure 4a). However, we cannot exclude the possibility that this reduction in sesquiterpene levels was partly due to a reduced plastidic IPP pool and its export to the cytosol, as observed in transgenic GPPS-SSU tobacco (Nicotiana tabacum) plants (Orlova et al., 2009).
Analysis of monoterpenes in these lines revealed that ripe GPPS-SSU × ZIS fruits produce comparable levels of geranial, neral and citronellol to those detected for fruits of the GPPS-SSU parental line, but these compounds were almost absent in the fruits of transgenic plants expressing ZIS alone (Figure 5a). In contrast, the levels of ZIS-derived monoterpenes, including α–thujene, α–pinene, sabinene and γ–terpinene, were significantly higher in GPPS-SSU × ZIS fruits than in ZIS fruits (Figure 5b,c and Table S3), despite an 86% decrease in ZIS mRNA abundance in the GPPS-SSU × ZIS cross. The increases in these monoterpenes, both those collected as emitted volatiles and directly extracted from fruits, ranged on average from 1.9- to 7.3-fold (Figure 5c and Table S3) and 2.1- to 4.6-fold (Figure 5d and Table S3), respectively, compared to fruits of the ZIS parental line. These results suggest that the increased GPP pool produced by over-expression of GPPS-SSU in plastids is available not only to monoterpene synthases (i.e. GES) present in the same subcellular compartment, but also to cytosolically localized terpene synthases, such as ZIS, which are able to use GPP as a substrate to form monoterpenes. To biochemically confirm the cytosolic localization of ZIS and the plastidic localization of GPPS, chromoplasts were isolated from GPPS-SSU × ZIS fruits, and enzyme activities were analyzed in crude extracts, cytosolic and plastidic fractions together with marker enzymes for assessing subcellular enrichment (Table 2). ZIS activity was found exclusively in the cytosol, while GPPS activity was detected in both plastids and cytosol. GPPS activity showed the same level of enrichment in the plastidic and cytosolic fractions (1.82- and 0.52-fold, respectively) as the plastidic marker NADP-linked glyceraldehyde 3–phosphate dehydrogenase (GAPDH; 1.98- and 0.62-fold, respectively), indicating that GPPS activity exists in plastids and that the activity observed in the cytosolic fraction derives from plastids that are broken during fractionation.
Table 2. Subcellular fractionation of crude extracts from GPPS-SSU × ZIS tomato fruits
Activities of marker enzymes, ZIS and GPPS were assayed in crude extracts, cytosolic and plastid fractions from tomato fruits. Alcohol dehydrogenase (ADH) and glyceraldehyde 3–phosphate dehydrogenase (GAPDH) were used as marker enzymes for the cytosol and plastids, respectively. Data are means ± SE (n = 3). ND, not detected.
GPPS-SSU over-expression provides substrate for monoterpene formation in plastids
GPPS is the key branchpoint enzyme leading to monoterpene biosynthesis, and is generally localized in plastids (Tholl et al., 2004). We hypothesized that an increase in plastidic GPPS activity via over-expression of the corresponding prenyltransferase in transgenic plants would re-direct the metabolic flux of MEP pathway-derived IPP and DMAPP towards GPP formation, and thus enhance the availability of substrate for monoterpene biosynthesis. The catalytically inactive GPPS-SSU of heteromeric snapdragon GPPS was previously found to interact with phyllogenetically distant GGPPSs and change their product specificity from GGPP to GPP (Tholl et al., 2004; Orlova et al., 2009). Over-expression of snapdragon GPPS-SSU in transgenic tobacco plants significantly increased GPPS activities in chloroplasts, as well as the levels of monoterpenes emitted from leaves and flowers compared to control plants (Orlova et al., 2009), indicating that the GPP pool available to endogenous monoterpene synthases limited overall monoterpene production. Unfortunately, GPPS-SSU tobacco lines displayed severe leaf chlorosis and dwarfism, probably due to reduced synthesis of GGPP required for production of chlorophyll, carotenoids and gibberellic acid (Orlova et al., 2009).
Here, snapdragon GPPS-SSU was over-expressed in transgenic tomato under the control of the PG promoter in order to restrict expression to ripening fruits. In ripening wild-type tomato fruits, the plastidic MEP pathway is highly active to provide GGPP precursors for the massive accumulation of lycopene and other carotenoids (Lawrence et al., 1997; Lois et al., 2000; Botella-Pavia et al., 2004; Paetzold et al., 2010). However, only minute amounts of monoterpenes are formed (Buttery et al., 1990; Baldwin et al., 2000; Lewinsohn et al., 2001; Davidovich-Rikanati et al., 2007) due to the limited number of endogenous monoterpene synthases expressed in fruits (Bleeker et al., 2011; Falara et al., 2011). GPPS-SSU over-expression in transgenic tomato plants had no obvious deleterious effects on vegetative tissues and general plant development, while lycopene levels in ripe fruits were reduced by up to 90% in some GPPS-SSU lines compared to controls (Figure 1c, and Figures S2b and S3b). In contrast, fruits of transgenic tomato plants expressing GES (Davidovich-Rikanati et al., 2007) showed only a 50% reduction in lycopene levels despite even higher monoterpene production (Figure 3b–d). Therefore, the severe reduction of lycopene levels in GPPS-SSU fruits cannot be attributed solely to a re-direction of IPP/DMAPP metabolic flux from GGPP to GPP formation. The change in GGPPS product specificity via the interaction with introduced snapdragon GPPS-SSU may also contribute to the reduction of lycopene formation as shown in tobacco (Orlova et al., 2009). However, in contrast to GPPS-SSU-expressing tobacco plants, GGPPS activity in GPPS-SSU tomato fruits was not reduced (Table 1), probably due to up-regulation of GGPPS–2 expression (Figure S1). To date, the reason for the reduction in the lycopene level remains unknown, but it occurs post-transcriptionally (Figure S1), probably via feed-forward regulation of lycopene biosynthetic enzymes by one (or several) products formed in transgenic lines. Indeed, phytoene synthase from Capsicum annum was shown to be inhibited by inorganic pyrophosphate, one of the products of the GPPS-catalyzed reaction (Dogbo et al., 1988). In addition, unaltered levels of β–carotene, lutein and zeaxanthin in GPPS-SSU-expressing lines suggest a complex regulation of carotenoid biosynthesis, which ensures the homeostasis of certain carotenoids.
The production of monoterpenes in ripe fruits of the GPPS-SSU tomato lines, both as emitted volatiles and endogenous internal pools (Figure 2 and Table S2), and their positive correlation with GPPS-SSU expression (Figure 1a,b), provides evidence for an increased plastidic GPP pool in transgenic fruits. However, the monoterpenes produced were not the result of action of endogenous monoterpene synthases, but of endogenous phosphatases (Perez et al., 1980; Izumi et al., 1996; Ganjewala and Luthra, 2009), reductase(s) and alcohol dehydrogenase(s) (Bicsak et al., 1982; Davidovich-Rikanati et al., 2007), which were previously shown to be present in tomato fruits (Davidovich-Rikanati et al., 2007). Geraniol and linalool were the products of enzymatic hydrolysis of GPP by phosphatases. Geraniol was subsequently converted to citronellol by a reductase similar to that found in rose (Rosa x hybrida) (Dunphy and Allcock, 1972). Both geraniol and citronellol were probably further oxidized to the monoterpene aldehydes geranial (which reversibly isomerizes to neral) and citronellal, respectively, by alcohol dehydrogenase(s) as described in several plant species (Potty and Bruemmer, 1970; Bicsak et al., 1982; Singh Sangwan et al., 1993; Sekiwa-Iijima et al., 2001; Luan et al., 2005; Iijima et al., 2006).
Over-expression of the monoterpene synthase gene GES in the GPPS-SSU background (GPPS-SSU × GES transgenic line) resulted in increased production of geraniol and its derivatives relative to parental GPPS-SSU and GES lines (Figure 3 and Table S2), suggesting that the increased plastidic GPP pool was available to GES for monoterpene formation. Interestingly, only the total amounts of monoterpenes detected in the headspace of GPPS-SSU × GES fruits were increased (Figure 3c and Table S2), but the internal monoterpene pools remained unchanged (Figure 3d and Table S2) compared to GES fruits. These results suggest that the overall metabolic flux through the engineered pathway was increased, with the internal monoterpene pools being saturated.
There are numerous successful examples of metabolic engineering via over-expressing plastidic monoterpene synthases in transgenic plants (Dudareva and Pichersky, 2008); however, to date a co-expression approach has only been used a few times. Wu et al. (2006) observed a roughly fivefold higher monoterpene accumulation by co-expressing a plastid-targeted GPPS and a lemon (Citrus limon) limonene synthase gene in transgenic tobacco plants compared with plants expressing the monoterpene synthase alone. Likewise, co-expression of a FPPS with a sesquiterpene synthase in tobacco lead to a drastic increase in product formation compared to plants only expressing the latter enzyme (Wu et al., 2006). The data presented here, as well as those reported by Wu et al. (2006), demonstrate that simultaneous expression of prenyltransferases and TPSs results in significantly higher formation of terpenes compared to expression of either of the two enzymes alone.
The increased plastidic GPP pool supports monoterpene formation in the cytosol
Recently, transgenic plants expressing a sesquiterpene synthase with monoterpene synthase activity (Davidovich-Rikanati et al., 2008), and monoterpene synthases lacking their plastid-targeting transit peptide (Ohara et al., 2003; Wu et al., 2006), were found to produce some monoterpenes, suggesting the presence of a small GPP pool in the cytosol. Moreover, a cytosolic terpene synthase in cultivated strawberry (FaNES1) was shown to use GPP as well as FPP as substrates to produce roughly similar amounts of the monoterpene linalool and the sesquiterpene nerolidol, respectively (Aharoni et al., 2004). While these discoveries further support the existence of a cytosolic GPP pool, they raise questions about the origin of GPP in this subcellular compartment.
GPPSs, irrespective of their homodimeric or heterodimeric architecture, are generally considered to be plastid-localized. The only known cytosolic GPPS exists in Lithospermum erythrorhizon and is involved in MVA pathway-dependent formation of the hemiterpenoid shikonin (Sommer et al., 1995; Li et al., 1998). Even though GPP represents an intermediate in the biosynthesis of FPP catalyzed by cytosolic FPPS, it was assumed that GPP stays bound to the active site of the enzyme. However, small amounts of GPP have been observed as a product in in vitro assays using recombinant FPPSs (Burke and Croteau, 2002; Hemmerlin et al., 2003b). Moreover, it has been hypothesized that the product specificity of FPPSs may be influenced by the relative concentrations of IPP and DMAPP substrates, with lower IPP concentrations leading to increased GPP formation (Szkopinska and Plochocka, 2005).
Despite the fact we cannot exclude the possibility that some cytosolic GPP may be formed as a side product of the reaction catalyzed by FPPS, our data obtained upon co-expression of the plastidic GPPS-SSU with the cytosolic terpene synthase ZIS suggest a different scenario for the origin of the cytosolic GPP pool. While transgenic tomato lines expressing ZIS alone accumulated high levels of α–zingiberene, as well as several other sesquiterpenes, in ripening fruits, they also contained small amounts of monoterpenes, including α–thujene, α–pinene, sabinene and γ–terpinene (Figure 5; Davidovich-Rikanati et al., 2008), consistent with the ability of ZIS to use GPP as a substrate. However, the total amount of ZIS-derived monoterpenes in ripe transgenic tomato fruits was increased by more than 3.5-fold when cytosolic ZIS was co-expressed with plastidic GPPS-SSU (Figure 5 and Table S3). The observed increase in cytosolic monoterpene production is even more remarkable considering that ZIS expression was lower in the GPPS-SSU × ZIS fruits compared to the parental ZIS line (roughly one-fifth, Figure 4a). This unprecedented finding suggests that (i) the increased GPP pool in plastids due to GPPS-SSU over-expression also supports monoterpene formation in the cytosol, and (ii) GPP export from plastids to the cytosol occurs.
To date, ample evidence for metabolic exchange between MVA pathway-derived and MEP pathway-derived IPP pools in the cytosol and plastids has been collected. In particular, IPP export from plastids (Figure 6) was shown to be of importance for cytosolic sesquiterpene synthesis in several plants (Steliopoulos et al., 2002; Dudareva et al., 2005; Hampel et al., 2005a,b; Orlova et al., 2009). In addition, transporter-mediated trafficking of IPP across the inner envelope membrane of plastids has been demonstrated using isolated organelles as well as isolated plastid envelope proteins reconstituted into liposomes (Soler et al., 1993; Bick and Lange, 2003; Flügge and Gao, 2005). However, substrate competition studies using proteoliposomes containing reconstituted inner envelope membrane proteins from spinach (Spinacia oleracea) chloroplast showed that this as yet unidentified transporter not only accepts IPP as a substrate, but also GPP, FPP and DMAPP (Bick and Lange, 2003). Analysis of its kinetic properties using the same system revealed that IPP and GPP are the preferred substrates for this unknown transporter, displaying similar Km and Vmax values (Bick and Lange, 2003). Our data support these in vitro assays, providing in vivo proof of GPP export from plastids to the cytosol for monoterpene synthesis (Figure 6).
The GPP export from plastids, and, as a consequence, the existence of a cytosolic GPP pool, shed new light on the function of cytosolic sesquiterpene synthases. To date, a diverse set of sesquiterpene synthases from various plant species, including gymnosperms and angiosperms (monocots and dicots), appears to be active with GPP in addition to FPP, and thus are able to produce monoterpenes (see Table S1). However, in the absence of a systematic characterization of all known plant sesquiterpene synthases with respect to their ability to use GPP, it is difficult to evaluate how widespread such cytosolic monoterpene synthase activity is. Thus, it should be considered that some monoterpenes produced in plant tissues may be synthesized by cytosolic (sesqui)terpene synthases using plastid-derived GPP rather than by plastidic monoterpene synthases. Stable isotope labeling experiments using 1–[1–13C]deoxy-d–xylulose and [1–13C]glucose with subsequent analysis of labeling patterns and isotope abundances revealed that, in chamomile (Matricaria recutita) sesquiterpenes, the first two isoprene units are predominantly formed via the MEP pathway, while the third unit has mixed MVA/MEP origin, suggesting incorporation of plastidic GPP into cytosolically synthesized FPP and sesquiterpenes (Adam and Zapp, 1998; Adam et al., 1999). Therefore, it is possible that export of MEP pathway-derived GPP from the plastids may also support cytosolic sesquiterpene synthesis in some plants.
Plant material and growth conditions
Tomato (Solanum lycopersicum) line MP–1 (Barg et al., 1997) was used for generation of transgenic plants as well as a control for all analyses. Plants were grown under a 14 h photoperiod in standard greenhouse conditions. MP–1 plants, as well as transgenic GES and ZIS tomato lines (Davidovich-Rikanati et al., 2007, 2008), were initially grown from seeds. Subsequently, all transgenic lines (GPPS-SSU, GES and ZIS) were propagated by re-rooting of cuttings on soil.
Vector construction and plant transformation
The A. majus GPPS-SSU coding region was cloned into the binary vector pBIN19 (Fray et al., 1994), which contains the tomato fruit-specific PG promoter (4.8 kb) and PG terminator (1.8 kb), and the kanamycin-resistance marker gene NPTII, driven by the CaMV 35S promoter (Nicholass et al., 1995). The binary vector was introduced into Agrobacterium tumefaciens strain c58C1/pMP90 (Koncz and Schell, 1986), which was used for transformation. Tomato MP–1 plants were transformed as described previously (McCormick et al., 1986) using kanamycin (100 mg L−1) for selection. Transgenic plants rooted on kanamycin-containing medium were transferred to soil and adapted to greenhouse conditions.
Molecular analysis of transgenic plants
Genomic DNA was extracted from tomato leaves as described previously (Ausubel et al., 1994). Tomato lines were screened for the presence of transgenes by PCR on genomic DNA using gene-specific primers for snapdragon GPPS-SSU and O. basilicum GES and ZIS (Table S4).
RNA isolation and quantitative real-time PCR
Total RNA was isolated from fruits of tomato plants as described previously (Eggermont et al., 1996). For quantitative real-time PCR analysis, total RNA was pre-treated with RNase-free DNase (Promega, www.promega.com) and cDNA was synthesized using Superscript II reverse transcriptase (Invitrogen, www.invitrogen.com). Gene-specific primers were designed using PrimerExpress software (Applied Biosystems, www.invitrogen.com) for snapdragon GPPS-SSU, O. basilicum GES and ZIS, and tomato DXS, DXR, GPPS-SSU-II, GGPPS–2, PPPS, PSY–1, PSY–2, PDS and Actin-Tom52 (Table S4). All primers showed more than 90% efficiency at final concentrations of 300 nm (Am-GPPS-SSU, Ob-GES, Sl-DXS, Sl-DXR, Sl-GPPS-SSU-II, Sl-PPPS, Sl-PSY–1 and Sl-Actin-Tom52), 500 nm (Sl-GGPPS–2, Sl-PSY–2 and Sl-PDS) and 900 nm (Ob-ZIS). Quantitative real-time PCR reactions were performed, and transcript levels were determined as described previously (Orlova et al., 2009) using the StepOne real-time PCR system (Applied Biosystems). Absolute transcript levels were determined for GPPS-SSU, GES and ZIS, and transcript levels of all tomato genes were analyzed relative to that of the reference gene Actin-Tom52. Each data point represents the mean of at least three independent biological samples with three technical replicates for each sample.
Protein extraction and immunoblot analysis
Crude protein extracts were prepared from fruits of MP–1 control and GPPS-SSU transgenic tomato plants as described previously (Dudareva et al., 1996). Total proteins (20 μg) were separated by SDS–PAGE, transferred to nitrocellulose membrane and probed using rabbit anti-GPPS-SSU polyclonal antibodies (Tholl et al., 2004; 1:2500 dilution). Antigen bands were visualized using goat anti-rabbit IgG horseradish peroxidase conjugate (Sigma-Aldrich, www.sigmaaldrich.com; 1:30 000 dilution) as secondary antibodies and chemiluminescence reagents (Perkin Elmer, www.perkinelmer.com) according to the manufacturer's instructions.
Crude protein extracts were prepared from Br + 3 tomato fruits in extraction buffer (1:2 w/v tissue/buffer) as described by Nagel et al. (2012), followed by desalting on 2 ml Zeba Spin columns (7 K molecular weight cut-off; Thermo Scientific, www.thermoscientific.com). Prenyltransferase assays were performed as described previously (Orlova et al., 2009) in the presence of 20 mm of the IPP isomerase inhibitor iodoacetamide (Sigma-Aldrich) using [1–14C]-IPP (55 mCi mmol−1) and DMAPP (for GPPS and FPPS activities) or FPP (for GGPPS activity). Each reaction containing 25 μg of protein was overlaid with 1 ml hexane and incubated for 30 min at 30°C. Assays were stopped by adding 3 N HCl, and additionally incubated for 20 min at 30°C. Hydrolysis products were extracted into the hexane phase, and aliquots were either counted in a liquid scintillation counter (GGPPS assays) or analyzed by TLC (GPPS/FPPS assays). For TLC analysis, 100 ng of geraniol, farnesol and gernylgeraniol were added to each sample as internal standard and carrier, and the hexane fractions were concentrated to 25 μl. The products were then separated on reversed-phase TLC plates (C18 silica TLC plates; Sorbent Technologies, www.sorbtech.com) using an acetone/water mobile phase (6:1 v/v). Radioactive spots were identified and quantified using a phosphor imager (Molecular Dynamics Typhoon 8600, www.gelifesciences.com). Internal standards were visualized by exposure to iodine vapor.
Measurement of carotenoid levels in tomato fruits
All sample handling, extraction and analysis were performed under yellow lights to minimize photo-oxidative reactions. Extraction of carotenoids from Br + 10 tomato fruits was performed as described previously (Goltz et al., 2012): 2.5 g of pericarp tissue were combined with 0.5 g sodium bicarbonate and 1 g Celite (Sigma Aldrich), and carotenoids were extracted using a 1:1 solution of acetone/petroleum ether (0.1% BHT) using a Polytron mixer (Kinematica, www.kinematica.ch). The suspension was filtered through Whatman (www.whatman.com) filter paper, and the remaining tissue homogenate was re-extracted three or four times. All acetone/petroleum ether fractions were combined and saponified using 40% w/w KOH in methanol. The mixture was transferred to a separatory funnel, and the ether layer was washed with 10 ml of saturated NaCl solution and distilled water. The petroleum ether phase was collected, poured through a column of sodium sulfate to remove residual water, and brought to a total volume of 100 ml. Aliquots (4 ml) of the petroleum ether fraction were dried under a stream of nitrogen, re-solubilized in a MeOH/ethyl acetate (1:1) mixture, and filtered through a 0.45 μm filter. Extracted carotenoids were then analyzed and quantified by HPLC as described previously (Kean et al., 2008).
Collection, extraction and analysis of terpenoid volatiles from tomato fruits
Emitted volatiles were collected from ripe tomato fruits (Br + 10 stage) using a closed-loop stripping method under growth chamber conditions (21°C, 50% relative humidity, 150 μmol m−2 per sec light intensity, and a 12 h photoperiod; Donath and Boland, 1995; Dudareva et al., 2005). Volatile collections were performed for 24 h using Porapak Q traps (80/100 mesh size; Alltech Associates, www.alltechweb.com), and analyzed as described previously (Dudareva et al., 2005). For analysis of internal pools of volatiles, approximately 30 g pericarp from fresh ripe tomato fruits (Br + 10) was cut into small pieces and extracted with 100 ml methyl tert-butyl ether as described previously (Davidovich-Rikanati et al., 2007). The ethereal phase was separated off, dried with anhydrous CaSO4, concentrated to 200 μl under gentle N2 flow, and supplemented with 3.33 μg of naphthalene as an internal standard. A 2 μl aliquot of the concentrated methyl tert-butyl ether extract was analyzed by GC–MS as described previously (Dudareva et al., 2005). Representative mono- and sesquiterpene standards were used to determine average response factors for both compound classes, which were used for quantification of the analyzed compounds.
Isolation of chromoplasts and analysis of ZIS, GPPS and marker enzyme activities
Chromoplasts were isolated as described by Barsan et al. (2010): 100 g pericarp from tomato fruits of the GPPS-SSU × ZIS line was blended, and the resulting homogenate was filtered through Miracloth (www.emdmillipore.com). After centrifugation at 2000 g for 5 min, the supernatant (cytosolic fraction) was recovered. The plastid pellet was resuspended in buffer (50 mm HEPES, 330 mm sorbitol, 1 mm MgCl2, 1 mm MnCl2, 2 mm EDTA), and loaded on a Percoll (GE Healthcare Life Sciences, www.gelifesciences.com) step gradient as described by Siddique et al. (2006). After centrifugation at 7000 g for 30 min, intact plastids were recovered, washed in buffer and used for further analyses. All steps of plastid isolation and fractionation were performed at 4°C. Plastids were osmotically lysed, and the stromal fraction was recovered by centrifugation at 29 000 g for 15 min. Prior to enzyme assays, all obtained fractions were desalted. ZIS activities were analyzed as described by Davidovich-Rikanati et al. (2008) with 30 μm FPP, and volatile products were extracted by solid-phase micro-extraction and analyzed by GC-MS. GPPS activities were analyzed as described previously (Orlova et al., 2009) using [1–14C]-IPP (55 mCi mmol−1). Enrichment of cytosolic and plastidic fractions was assessed by measuring the activities of marker enzymes: cytosolic alcohol dehydrogenase (ADH; Smith and ap Rees, 1979) and plastidic NADP-linked glyceraldehyde 3–phosphate dehydrogenase (GAPDH; Quinlivan et al., 2003; Oostende et al., 2008).
This work was supported by Agricultural and Food Research Initiative competitive grant number 2008-35318-04541 to N.D. and E.P., and by United States/Israel Bi-national Agriculture Research and Development grant number IS-4125-08C to N.D., Y.S., E.P. and E.L. We thank Milena Leon-Garcia (Department of Food Science, Purdue University) for technical assistance with carotenoid analysis.