Improved fruit α‐tocopherol, carotenoid, squalene and phytosterol contents through manipulation of Brassica juncea 3‐HYDROXY‐3‐METHYLGLUTARYL‐COA SYNTHASE1 in transgenic tomato

Summary 3‐Hydroxy‐3‐methylglutaryl‐coenzyme A synthase (HMGS) in the mevalonate (MVA) pathway generates isoprenoids including phytosterols. Dietary phytosterols are important because they can lower blood cholesterol levels. Previously, the overexpression of Brassica juncea wild‐type (wt) and mutant (S359A) BjHMGS1 in Arabidopsis up‐regulated several genes in sterol biosynthesis and increased sterol content. Recombinant S359A had earlier displayed a 10‐fold higher in vitro enzyme activity. Furthermore, tobacco HMGS overexpressors (OEs) exhibited improved sterol content, plant growth and seed yield. Increased growth and seed yield in tobacco OE‐S359A over OE‐wtBjHMGS1 coincided with elevations in NtSQS expression and sterol content. Herein, the overexpression of wt and mutant (S359A) BjHMGS1 in a crop plant, tomato (Solanum lycopersicum), caused an accumulation of MVA‐derived squalene and phytosterols, as well as methylerythritol phosphate (MEP)‐derived α‐tocopherol (vitamin E) and carotenoids, which are important to human health as antioxidants. In tomato HMGS‐OE seedlings, genes associated with the biosyntheses of C10, C15 and C20 universal precursors of isoprenoids, phytosterols, brassinosteroids, dolichols, methylerythritol phosphate, carotenoid and vitamin E were up‐regulated. In OE‐S359A tomato fruits, increased squalene and phytosterol contents over OE‐wtBjHMGS1 were attributed to heightened SlHMGR2, SlFPS1, SlSQS and SlCYP710A11 expression. In both tomato OE‐wtBjHMGS1 and OE‐S359A fruits, the up‐regulation of SlGPS and SlGGPPS1 in the MEP pathway that led to α‐tocopherol and carotenoid accumulation indicated cross‐talk between the MVA and MEP pathways. Taken together, the manipulation of BjHMGS1 represents a promising strategy to simultaneously elevate health‐promoting squalene, phytosterols, α‐tocopherol and carotenoids in tomato, an edible fruit.

Given that dietary phytosterols have been reported to lower blood cholesterol levels and might thereby reduce the risk of heart disease (Bradford and Awad, 2007;Moreau et al., 2002;Woyengo et al., 2009), it would be strategic to genetically manipulate HMGS in an edible fruit such as tomato. The metabolic engineering of the tomato MVA pathway using HMGR had caused a 2.4-fold increase in phytosterol content in mature transgenic T 0 tomato fruits overexpressing Arabidopsis HMGR1 (HMGR1-OE) (Enfissi et al., 2005). However, in the mature fruits of the homozygous T 2 tomato HMGR1-OE, total HMGR activity was not enhanced and only few specific phytosterols were elevated (Enfissi et al., 2005), indicating that phytosterol increase in tomato HMGR1-OEs was not stably inherited (Enfissi et al., 2005). Furthermore, HMGR is known to be subject to regulation at transcriptional, post-transcriptional, translational and posttranslational levels (Bach, 1986;Hemmerlin, 2013;Wong et al., 1982).
Given the positive effects of BjHMGS1 in elevating phytosterols in HMGS-OEs of model plants such as Arabidopsis and tobacco, and the improved effect of OE-S359A in plant growth besides sterol content, the application of S359A should be extended to benefit food crops. Tomato (Solanum lycopersicum) was selected in this study because it is a popular fruit, eaten raw and from which juice is easily extracted for human consumption. Furthermore, the technology for tomato transformation is available (Mathews et al., 2003). To comprehensively investigate the effects of HMGS in regulating isoprenoid biosynthesis in tomato, the expression of genes in tomato HMGS-OEs involved in the biosyntheses of MVA, C10, C15 and C20 universal precursors of isoprenoids, phytosterols, BR, CKs, dolichols, monoterpenes, sesquiterpenes, MEP, carotenoids and vitamin E was analysed by quantitative reverse transcription PCR (qRT-PCR). The effect of HMGS overexpression on the metabolic flux was subsequently investigated by gas chromatography-mass spectrometry (GC-MS) and high-performance liquid chromatography (HPLC). It is interesting to note that the overexpression of BjHMGS1 in tomato upregulated SlGPS and SlGGPPS, causing a significant elevation in fruit a-tocopherol and carotenoids, products of the MEP pathway, besides those of the MVA pathway such as squalene and phytosterols. Overall, this study demonstrates the potential of BjHMGS1 in simultaneously promoting vitamin E, carotenoid, squalene and phytosterol production in edible plants that would benefit human health.

Tomato HMGS-OE plants showed increased growth
To check whether BjHMGS1 overexpression in verified tomato lines (Figures 1, S1 and S2) caused enhanced growth as previously observed in tobacco HMGS-OEs (Liao et al., 2014b), the length and fresh weight of 3-week-old (Figure 2a-c) and 5-week-old (Figure 2d and e) seedlings and the height of 9-weekold tomato plants (Figure 2f and g) were compared amongst OE-wtBjHMGS1, OE-S359A and the vector-transformed lines. Three-week-old OE-wtBjHMGS1 and OE-S359A tomato seedlings had a higher fresh weight (83% and 57%, respectively, in Figure 2a and b) and longer seedling length (16% and 19%, respectively, in Figure 2a and c) than the control, but there was no significant difference in growth between 3-week-old OE-wtBjHMGS1 and OE-S359A seedlings (Figure 2b and c). Correspondingly, 5-week-old transgenic OE-wtBjHMGS1 and OE-S359A tomato plants displayed significant increases (17% and 26%, respectively) in height over the control (Figure 2d and e). Consistently, 9-week-old HMGS-OEs (OE-wtBjHMGS1 and OE-S359A) grew better than the control ( Figure 2f); OE-wtBjHMGS1 displayed a 22% increase in height over the control, while OE-S359A displayed an even higher (39%) increase as shown in Figure 2g.
It is interesting to note that carotenoid-related genes [(PHY- and vitamin E-related genes (SlGGPPR and SlGMTT) were up-regulated in tomato HMGS-OE seedlings ( Figure S6a and b). Furthermore, SlPDS, SlCRTISO and SlCYC-B expression in OE-S359A was much higher than in OE-wtBjHMGS1 ( Figure S6a). However, there was no difference in monoterpene-related SlMTS1 expression amongst the vector control, OE-wtBjHMGS1 and OE-S359A ( Figure S5b).

Tomato HMGS-OE fruits accumulated MVA-derived squalene, sterol-related intermediates, sterols and MEPderived vitamin E and carotenoids
Given the induced expression of isoprenoid-, vitamin E-and carotenoid-related genes in tomato HMGS-OE seedlings and tomato fruit is edible, analysis on tomato HMGS-OE fruits would inform on feasibility in the accumulation of health-promoting end products from the MVA and MEP pathways such as phytosterols and related intermediates, vitamin E and carotenoids.
When the molecular mechanism for the accumulation of MVAderived compounds in HMGS-OE fruits was investigated, the expression of SlFPS1, SlGPS, SlSQS, SlSQE, SlCAS1, SlCYP85A1 and SlCYP85A3 was found to be significantly higher in OE-wtBjHMGS1 and OE-S359A than in the control, with the exception of SlSQS in OE-wtBjHMGS1 line 430 and SlCYP85A1 in OE-S359A line 625 ( Figure S7a and b). SlFPS1, SlGPS, SlSQS and SlCYP710A11 expression in OE-S359A was much higher than in OE-BjHMGS1 ( Figure S7a and b). Conversely, the expression of SlHMGR1, SlHMGR2 and SlIPI was down-regulated in HMGS-OE fruits ( Figure S7a).
More interestingly, a significant increase in vitamin E (a-tocopherol) in tomato HMGS-OE fruit lipid extracts was noted ( Figure 5) from GC-MS analysis. Vitamin E content in OE-BjHMGS1 and OE-S359A was significantly higher than in the vector control ( Figure 5). Its content in HMGS-OEs increased to 138.6-187.5 lg/g dry weight, in comparison with 27.5 lg/g dry weight in the control ( Figure 5), representing average increases of 521% in OE-BjHMGS1 and 494% in OE-S359A ( Figure 5). In HPLC analysis, a significant enhancement of carotenoids (lycopene and b-carotene) in tomato HMGS-OE fruit was observed ( Figure 6). Lycopene and b-carotene contents in OE-BjHMGS1 and OE-S359A were higher than in the control ( Figure 6, Tables 1 and 2). Lycopene content in HMGS-OEs increased to 6.0-10.7 mg/g dry weight, in comparison with 3.0 mg/g dry weight in the control (Table 1), representing average elevations of 234% in OE-BjHMGS1 and 111% in OE-S359A (Table 2). b-Carotene content in HMGS-OEs went up to 3.5-6.6 mg/g dry weight, in comparison with 1.5 mg/g dry weight in the control (Table 1), representing average increases of 247% in OE-BjHMGS1 and 169% in OE-S359A (Table 2). However, lycopene and total carotenoid contents in OE-S359A were significantly lower than in OE-BjHMGS1 ( Figure 6, Tables 1 and 2). Lycopene and total carotenoid contents decreased by 53% and 46%, respectively, in OE-S359A than in OE-wtBjHMGS1 ( Table 2). The elevation in vitamin E and carotenoids in OE-BjHMGS1 and OE-S359A fruits ( Figure 5) coincided with an up-regulation of plastidial SlGPS and SlGGPPS1, but not of SlGGPPS2 in OE-wtBjHMGS1 and OE-S359A fruits ( Figure S7c).    (Tables 1 and 2), they were tested for antioxidant activity in comparison with the vector control. Tomato HMGS-OE fruits displayed significantly (P < 0.05) higher DPPH (1,1-diphenyl-2picrylhydrazyl) radical scavenging activities than the control (Figure 7). OE-wtBjHMGS1 showed 257.9%-299.9% higher antioxidant activity than the control, while OE-S359A was 89.5%-96.5% higher. OE-wtBjHMGS1 possessed 82.1%-111.2% higher antioxidant activity than OE-S359A (Figure 7), and this corresponded to a higher carotenoid content in OE-wtBjHMGS1 than in OE-S359A (Tables 1 and 2).

Discussion
Elevation of health-promoting components achieved via metabolic engineering of BjHMGS1 in tomato fruit Resembling tobacco HMGS-OEs, tomato HMGS-OEs showed an enhanced expression of SlHMGR1, SlSQS, SlGGPPS1 and SlCYP85A1 and an increase in growth accompanied by a decline in SlHMGS2 and SlIPI expression. Furthermore, sterol content was elevated in HMGS-OE tomato fruits, confirming that elevation of health-promoting components was achieved via metabolic engineering of BjHMGS1 in tomato fruit and that extended phytosterol enhancement in fruits was attained using mutant S359A. New observations on tomato HMGS-OE seedlings and fruits not previously reported for tobacco HMGS-OE seedlings and leaves included: (i) Up-regulation of genes associated with the biosyntheses of C10, C15 and C20 universal precursors of isoprenoids, phytosterols and dolichols in seedlings; (ii) Up-regulation of MEP-, carotenoid-and vitamin E-related genes, but down-regulation of carotenoid-related gene SlPDS and no apparent effect on monoterpene-related gene (SlMTS1) expression in seedlings; (iii) Greater expression of MVA-related genes, genes leading to C10, C15, C20 universal precursors of isoprenoids and sesquiterpene-related genes in OE-S359A than that of OE-wtBjHMGS1  seedlings, corresponding to significant growth enhancement in 9-week-old OE-S359A plants, suggesting that S359A overexpression affected the expression of these isoprenoid precursor-and sesquiterpene-related genes; (iv) Up-regulation of SlFPS1, SlSQE, SlCAS1, SlCYP85A1, SlCYP85A3, SlGPS and SlGGPPS1 in HMGS-OE fruits, resulting in higher amounts of health-promoting components including squalene, vitamin E (a-tocopherol) and carotenoids; and (v) Elevation of fruit squalene and phytosterols in OE-S359A in comparison with OE-wtBjHMGS1, which was attributed to higher expression of SlHMGR2, SlFPS1, SlGPS, SlSQS and SlCYP710A11 in OE-S359A fruits.

HMGS regulates isoprenoid biosynthesis genes in tomato HMGS-OE seedlings
It has been reported that NtHMGR1 is a housekeeping gene, while NtHMGR2 is stress-inducible (Hemmerlin et al., 2004;Merret et al., 2007). This corresponds well to the up-regulation of NtHMGR1, but not NtHMGR2, in tobacco HMGS-OE seedlings (Liao et al., 2014b). In tomato, SlHMGR1 is known to be highly expressed at the early stage of fruit development, while SlHMGR2 is confined to fruit maturation and ripening (Narita and Gruissem, 1989). Not surprisingly, SlHMGR1, but not SlHMGR2, expression closely coincided with growth-dependent phytosterol biosynthesis (Rodr ıguez-Concepci on and Gruissem, 1999). In this study, the differential expression of SlHMGR1 and SlHMGR2 in tomato seedlings ( Figure 3) resembles that of tomato fruits (Narita and Gruissem, 1989; Rodr ıguez-Concepci on and Gruissem, 1999). Besides SlHMGR2, SlIPI was significantly down-regulated in tomato HMGS-OE seedlings (Figure 3), like NtIPI1 in transgenic tobacco HMGS-OEs (Liao et al., 2014b). Besides IPI, plastidlocalized SlCPT1 and SlCPT2 prefer DMAPP as a substrate (Akhtar et al., 2013), and their mRNA expression in tomato OE-S359A seedlings was significantly higher than in OE-wtBjHMGS1 (Figure 3), corresponding to better growth ( Figure 2). Interestingly, the increase in SlCPT1 expression was greater than that of SlCPT2 in these tomato HMGS-OE seedlings ( Figure S5a), implying that HMGS overexpression exerted a stronger effect on SlCPT1.
SlFPS1 has been proposed to play an important role in early fruit development as well as cell division and elongation (Gaffe et al., 2000). Simultaneous silencing of both FPS in Arabidopsis reduced sterol content and retarded seedling growth . While NtFPPS expression was slightly elevated in tobacco seedlings (Liao et al., 2014b), SlFPS1 expression in tomato OE-wtBjHMGS1 seedlings remained unchanged and was up-regulated in OE-S359A (Figure 3). Higher SlFPS1 expression observed in OE-S359A in comparison with OE-wtBjHMGS1 tomato seedlings (Figure 3) coincided well with a positive growth effect in 9-week-old OE-S359A, suggesting that S359A overexpression was more effective in enhancing SlFPS1 expression.
Squalene, product of SQS (Abe et al., 1993;Devarenne et al., 1998Devarenne et al., , 2002Seo et al., 2005), has been reported to confer health benefits because it possesses antitumour properties (Mathews, 1992;Newmark, 1997;Smith, 2000), and abilities in quenching (b-f) HPLC chromatograms of carotenoids from pSa13, OE-wtBjHMGS1 (430 and 450) and OE-S359A (622 and 625), respectively. Peak 1, lycopene; peak 2, b-carotene. The enlarged HPLC chromatogram of peak 2 and photograph of carotenoid extract dissolved in chloroform from each sample is presented in b-f. singlet oxygen (Kohno et al., 1995) and reducing phenobarbital, theophylline and strychnine in animals (Kamimura et al., 1992). The overexpression of Panax ginseng SQS1 enhanced sterol and ginsenoside contents in transgenic P. ginseng roots (Lee et al., 2004). Transgenic Arabidopsis overexpressing Glycine max SQS1 demonstrated significant elevation in seed sterols (Nguyen et al., 2013). The silencing of Withania somnifera SQS down-regulated downstream sterol pathway genes, reduced squalene, and sterol content and caused a dwarf phenotype (Singh et al., 2015). In this study, an enhanced effect of S359A on squalene production ( Figure 5, Tables 1 and 2) mirrored the benefits seen in P. ginseng and Arabidopsis (Lee et al., 2004;Nguyen et al., 2013). The increase in SlSQS expression correlated with enhanced squalene and sterol accumulation in tomato HMGS-OE fruits and higher SlSQS expression in OE-S359A tomato seedlings and fruits corresponded well with elevated squalene and sterol contents, consistent with greater expression of NtSQS and AtSQS (Liao et al., 2014b;Wang et al., 2012). However, reports of a dwarf phenotype from the overexpression of a truncated yeast SQS or codon-optimized Flag-tagged yeast SQS in tobacco (Pasoreck et al., 2016;Wu et al., 2012) suggest differences do occur between applications of plant and yeast SQS.
SlGGPPS1 was highly expressed in tomato leaves in contrast to SlGGPPS2, which was induced in fruits and flowers (Ament et al., 2006). Both SlGGPPS1 and SlGGPPS2 expression was upregulated in tomato HMGS-OE seedlings ( Figure S5), but only NtGGPPS2 was elevated in several lines of tobacco HMGS-OE seedlings, while NtGGPPS1, NtGGPPS3 and NtGGPPS4 remained down-regulated in all previous lines tested (Liao et al., 2014b). These results suggest that HMGS overexpression produced inconsistent effects on GGPPS expression in tomato and tobacco seedlings. It appears that HMGS overexpression caused a stronger positive effect on GGPPS expression in tomato seedlings in comparison with tobacco. Ruiz-Sola et al. (2016) identified one GGPPS gene encoding two differentially targeted (plastidial and cytosol) GGPPSs in Arabidopsis, implying that such could also occur in tomato and tobacco.

Co-up-regulation of BR-, CK-and dolichol-related genes in tomato HMGS-OEs
In higher plants, besides sterols, BRs and CKs are essential in growth and development (He et al., 2003;Howell et al., 2003;Li et al., 1996;Shani et al., 2010;Vriet et al., 2012;Wang et al., 2012), while dolichol is important in protein glycosylation . BR-related genes were up-regulated in Arabidopsis and tobacco HMGS-OEs (Liao et al., 2014b;Wang et al., 2012). A BR (SlCYP85A3)-and a dolichol (SlCPT3)-related gene were dramatically up-regulated in tomato HMGS-OE seedlings and more highly expressed in OE-S359A than in OE-wtBjHMGS1 ( Figure S4a and c), indicating an effect from HMGS overexpression. In Caenorhabditis elegans, HMGS is significant in the miRNA pathway by regulating the function of many miRNAs during development (Shi and Ruvkun, 2012). MVA-derived dolichols, which are involved in N-glycosylation, are essential for the activity of miRNAs in silencing their target mRNAs (Shi and Ruvkun, 2012). More experiments are needed to address the relationship between HMGS and dolichol accumulation in plants.

The overexpression of cytosolic HMGS promotes carotenoid and vitamin E formation in plastids
The overexpression of Arabidopsis HMGR in Lavandula latifolia and that of cytosolic isopentenyl phosphate kinase in Arabidopsis and tobacco increased both MVA-derived sterols and MEPderived monoterpenes and sesquiterpenes (Henry et al., 2015;Muñoz-Bertomeu et al., 2007). Salvia miltiorrhiza HMGR overexpression in hairy roots enhanced MEP-associated diterpene tanshinone accumulation (Kai et al., 2011). In this study, MEPrelated genes (SlDXS1, SlDXS2 and SlDXR), most carotenoidrelated genes (SlPSY-1, SlPSY-2, SlZDS, SlCRTISO, SlLCY-B, SlCYC-B and SlLCY-E with the exception of SlPDS) and vitamin E-related genes (SlGGPPR and SlGMTT) were up-regulated in tomato HMGS-OE seedlings (Figures 4 and S6), suggesting that HMGS overexpression in the cytosol can affect the biosynthesis of plastidial MEP-related isoprenoids including carotenoids and vitamin E. Indeed, total carotenoids increased in tomato HMGS-OE fruits (Figure 6, Tables 1 and 2) and lycopene content and total carotenoids were much higher in OE-wtBjHMGS1 than in OE-S359A ( Figure 6, Tables 1 and 2). In contrast, the expression of SlCRTISO and SlCYC-B in tomato OE-S359A seedlings was significantly higher than in OE-wtBjHMGS1 ( Figure S6). These results implied that differential carotenoid-related gene expression had probably occurred between tomato HMGS-OE seedlings and fruits, or post-transcriptional/post-translational regulation may have taken effect. Furthermore, increased HMGS enzyme activity in OE-S359A resulted in enhanced MVA-derived squalene and phytosterols but not MEP-derived carotenoids and vitamin E (Tables 1 and 2).
Vitamin E consists of two forms, tocopherols and tocotrienols (Brigelius-Floh e and Traber, 1999). Previous studies manipulated enzymes in the vitamin E biosynthetic pathway to increase leaf (37%: 10-fold) and seed (18%-1500%) vitamin E content in model plants Arabidopsis and tobacco, as well as crop plants including canola, soybean, corn, lettuce, potato and sunflower (Chen et al., 2006 and references cited therein;Del Moral et al., 2013;DellaPenna, 2005 and references cited therein;Vom Dorp et al., 2015). Also, the overexpression of zeaxanthin epoxidase in potato resulted in a two-to threefold elevation in a-tocopherol (R€ omer et al., 2002). However, the metabolic engineering of the MVA pathway or HMGS for a-tocopherol accumulation in a fruit crop had not been reported. We successfully demonstrated herein that manipulation of HMGS from the MVA pathway in tomato led to dramatic increase (~5-fold) in fruit a-tocopherol ( Figure 5). This confirms that the overexpression of a cytosolic HMGS could cause an increase in plastidial GGPP-derived vitamin E. There is evidence of cross-talk between the MVA and MEP pathways in up-regulated SlGGPPS1 expression in HMGS-OE tomato fruits, promoted a-tocopherol production. It is worth noting that there was no significant difference in the level of a-tocopherol between OE-wtBjHMGS1 and OE-S359A, indicating that S359A was not superior to wtBjHMGS1 in enhancing a-tocopherol production. Hence, metabolic engineering of HMGS from the MVA pathway can provide an alternative strategy in elevating a-tocopherol production in a fruit crop. Also, the tomato HMGS-OE fruits generated in this study present potential as an emerging form of beneficial food enriched in healthpromoting components including squalene, phytosterols, carotenoids and a-tocopherol.

Plant materials and growth conditions
Wild-type tomato (Lycopersicon esculentum Mill. cv. UC82B) seeds were obtained from Dr. WK Yip, The University of Hong Kong. Tomato seeds were surface-sterilized in 75% ethanol for 1 min, rinsed thrice in sterilized water, soaked in 25% Clorox for 10 min and rinsed four times with sterilized water. Seeds were transferred to MS medium for 2 days at 4°C before being moved to a tissue culture room for germination and seedling development. Tomato plants were grown at 25°C (16-h light)/ 22°C (8-h dark).

Generation and characterization of transgenic tomato overexpressing HMGS
Plasmids pBj134 (wt-BjHMGS1) and pBj136 (S359A) from Wang et al. (2012) were used for Agrobacterium-mediated tomato transformation (Mathews et al., 2003) with vector control pSa13 (Xiao et al., 2008). T 1 transgenic tomato seeds were screened on MS with 50 lg/ml kanamycin and analysed by PCR followed by DNA sequencing (Liao et al., 2014b;Wang et al., 2012). T 2 homozygous lines with single copy of BjHMGS1/S359A were analysed in mRNA and protein expression, plant growth and metabolite composition. Tomato total protein was extracted (Chye et al., 1999) from 3-week-old fresh tomato leaves and protein concentration measured (Bradford, 1976). Western blot analysis was conducted as described previously (Liao et al., 2014b;Wang et al., 2012;Xiao et al., 2010). Antibodies against BjHMGS1 were used in Western blot analysis (Wang et al., 2012). EcoRI-digested tomato genomic DNA (40 lg) from 4-week-old leaves was separated on agarose gel (0.7%) by electrophoresis. Southern blot analysis (Southern, 2006) was performed using a digoxigenin-labelled full length of BjHMGS1 cDNA probe generated by primer pair ML264 and ML276 (Wang et al., 2012). Primers are listed in Table S1.

Semiquantitative reverse transcription PCR (RT-PCR)
Total RNA from 3-week-old tomato seedlings was extracted using RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) followed by DNase I treatment (Qiagen). First-strand cDNA was synthesized from 5 lg total RNA from 3-week-old tomato seedlings using the SuperScript First-Strand Synthesis System (Invitrogen, Carlsbad, CA, USA). Semiquantitative reverse transcription PCR (RT-PCR) was conducted using the PCR System (Bio-Rad, Hercules, USA) with BjHMGS1-specific primer pair (ML1666 and ML1667) and tomato ACTIN primer pair (ML1688 and ML1689). Tomato ACTIN (SlACTIN), which has been previously tested (EI-Sharkawy et al., 2016;Melilli et al., 2014), was used as an internal control to estimate the amount of RNA in each sample. The conditions for PCR were as follows: denaturation at 95°C for 5 min, followed by 35 cycles of amplification (95°C for 15 s, 60°C for 20 s and 72°C for 20 s) and extension at 72°C for 10 min. The experiment was repeated two times. Primers for RT-PCR are listed in Table S1.

Quantitative reverse transcription PCR
Total RNA from 3-week-old tomato seedlings or mature red tomato fruits aged 57 days after pollination (DAP; breaker+15) was extracted using RNeasy Plant Mini Kit (Qiagen). The RNA (5 lg) was treated using DNase I (Qiagen) before reversetranscribed into first-strand cDNA using the SuperScript First-Strand Synthesis System (Invitrogen). Quantitative reverse transcription PCR (qRT-PCR) was carried out with a StepOne Plus Real-time PCR System (Applied Biosystems, Foster City, CA, USA) and FastStart Universal SYBR Green Master (Roche, Mannheim, Germany). The conditions for qRT-PCR were as follows: denaturation at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Three experimental replicates for each reaction were carried out using gene-specific primers, and tomato ACTIN was used as the internal control. The comparative C T method was used to analyse the qRT-PCR data (Schmittgen and Livak, 2008). The relative expression was normalized to SlACTIN, and the relative mRNA level in each HMGS-OE line in comparison with the empty vector control from three independent experiments was presented on the y-axis. Significant differences in the relative mRNA levels between different samples were analysed by the Student's t-test. Primers for qRT-PCR are listed in Table S1.

Measurements of growth rate
Phenotypic changes in tomato plants were analysed (Johnston and Dore, 1929). Single-copy T 2 homozygous plants were compared in plant growth. Four-day-old tomato seedlings were moved to fresh MS plates and grown for 8 days. Twelve-day-old tomato seedlings of similar size were transferred to soil for further analysis; height measurements of 5-week-old and 9-week-old tomato plants were taken. For each OE construct, two independent lines were tested. Thirty plants per individual line were used for measurements in height.

Statistical analysis
Significant differences in data between different samples were analysed by the Student's t-test.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Figure S1 PCR analysis on representative transgenic tomato HMGS-OEs. Figure S2 Southern blot analysis of representative transgenic tomato HMGS-OEs. Figure S3 QRT-PCR analysis on the expression of sesquiterpenerelated genes in tomato HMGS-OE seedlings. Figure S4 QRT-PCR analysis on the expression of BR-, cytokininand dolichol-related genes in tomato HMGS-OE seedlings. Figure S5 QRT-PCR analysis on the expression of C10 and C20 universal precursors of isoprenoid-, and monoterpene-related genes in tomato HMGS-OE seedlings. Figure S6 QRT-PCR analysis on the expression of MEP-derived carotenoid-and vitamin E-related genes in tomato HMGS-OE seedlings. Figure S7 QRT-PCR analysis on the expression of genes downstream of HMGS and plastidial GGPPSs in tomato HMGS-OE fruits. Table S1 Oligonucleotide primers used in this study. Data S1 Supplementary Methods. Data S2 Supplementary Result.