Efficient metabolic pathway engineering in transgenic tobacco and tomato plastids with synthetic multigene operons

Abstract

The engineering of complex metabolic pathways requires the concerted expression of multiple genes. In plastids (chloroplasts) of plant cells, genes are organized in operons that are coexpressed as polycistronic transcripts and then often are processed further into monocistronic mRNAs. Here we have used the tocochromanol pathway (providing tocopherols and tocotrienols, collectively also referred to as “vitamin E”) as an example to establish principles of successful multigene engineering by stable transformation of the chloroplast genome, a technology not afflicted with epigenetic variation and/or instability of transgene expression. Testing a series of single-gene constructs (encoding homogentisate phytyltransferase, tocopherol cyclase, and γ-tocopherol methyltransferase) and rationally designed synthetic operons in tobacco and tomato plants, we (i) confirmed previous results suggesting homogentisate phytyltransferase as the limiting enzymatic step in the pathway, (ii) comparatively characterized the bottlenecks in tocopherol biosynthesis in transplastomic leaves and tomato fruits, and (iii) achieved an up to tenfold increase in total tocochromanol accumulation. In addition, our results uncovered an unexpected light-dependent regulatory link between tocochromanol metabolism and the pathways of photosynthetic pigment biosynthesis. The synthetic operon design developed here will facilitate future synthetic biology applications in plastids, especially the design of artificial operons that introduce novel biochemical pathways into plants.

Tocochromanols (tocopherols and tocotrienols) are products of plant metabolism that arise from two biochemical pathways: the shikimate pathway of aromatic amino acid biosynthesis and the isoprenoid biosynthetic pathway (Fig. 1). Tocopherols arise from a condensation reaction of homogentisic acid (HGA) with phytyl diphosphate (PDP), whereas tocotrienols are produced by geranylgeranyl diphosphate (GGDP) attachment to HGA (Fig. 1). Both tocopherols and tocotrienols occur in four natural forms that differ in their methylation patterns of the aromatic head group: α, β, γ, and δ. Starting with the condensation of HGA to the isoprenoid chain (PDP or GGDP), all steps of tocochromanol biosynthesis take place in the plastid. Drawing on the plastid isoprenoid pool, tocochromanol biosynthesis is highly connected to a number of other metabolic pathways in the plastid compartment, including the biosynthesis of carotenoids, chlorophylls, gibberellins, phylloquinone, and plastoquinone (Fig. 1).

Fig. 1.

Metabolic pathway of tocochromanol (tocopherol and tocotrienol) biosynthesis in higher plants and its links to other metabolic processes. Tocochromanol biosynthesis draws on metabolites from two biochemical pathways: the shikimate pathway and the 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway of isoprenoid metabolism. The key difference between tocopherol and tocotrienol biosynthesis lies in the nature of the side chain, which comes from PDP in the case of tocopherols and from GGDP in the case of tocotrienols. In tocopherol biosynthesis, shikimate-derived HGA is connected to PDP in a reaction catalyzed by HPT. The analogous reaction in tocotrienol biosynthesis is catalyzed by HGGT and links GGDP to HGA. α-, β-, γ-, and δ-tocochromanols differ only in the number and arrangement of methyl groups on the aromatic ring. Enzymes, compounds, and bonds specific to tocotrienol biosynthesis are indicated in green. MSBQ, 2-methyl-6-solanesyl-1,4-benzoquinol; MPBQ, 2-methyl-6-phytylbenzoquinone; MGGBQ, 2-methyl-6-geranylgeranylbenzoquinone; DMPBQ, 2,3-dimethyl-5-phytyl-1,4-benzoquinone; DMGGBQ, 2,3-dimethyl-5-geranylgeranyl-1,4-benzoquinone; SAM, S-adenosyl methionine.

Tocochromanols are best known for their vitamin E activity. Because animals (including humans) cannot synthesize vitamin E, they depend on the uptake of plant-derived tocochromanols in their diet. In addition to being an essential micronutrient, tocochromanols have been linked to a variety of health-promoting effects [e.g., in preventing chronic disease (1)]. Most of these beneficial properties generally are explained by tocochromanols being highly potent antioxidants and efficient quenchers of reactive oxygen species. This activity comes from the ability of tocochromanols to donate a hydrogen atom from the hydroxyl group in the chromanol ring to a free radical, followed by mesomeric stabilization of the molecule’s own unpaired electron. The current models of tocochromanol action in the body are based largely on their radical-scavenging properties, in that protection of membranes from lipid peroxidation and protection of active sites of metabolic enzymes and nucleic acids from damage by free radicals are believed to be the main beneficial effects of tocochromanols at the cellular level.

Because of the high nutritional value and the benefits of vitamin E in human health, increasing the tocochromanol content of major agricultural crops has long been in the focus of breeding programs and genetic engineering approaches (2, 3). Elevated vitamin E contents can be achieved either by directing more metabolites into the tocopherol biosynthetic pathway or, alternatively, by altering the tocopherol spectrum [e.g., by converting other tocopherol forms into α-tocopherol, the form with the highest vitamin E activity; (4)].

Relatively little is known about the regulation of tocochromanol biosynthesis and its interconnection with other metabolic pathways in the plant cell. Mutant analysis and overexpression studies in the model plant Arabidopsis thaliana have shed some light on limiting enzymatic steps and also have revealed some strategies for manipulating the tocopherol content and altering the tocopherol spectrum (4–6). Here we used stable transformation of the plastid (chloroplast) genome in the model plant tobacco and the food crop tomato to engineer the tocopherol metabolic pathway and assess the impact of the three plastid-localized enzymes of the pathway (Fig. 1) on tocochromanol biosynthesis in chloroplasts and chromoplasts. Our results confirm previous data on the limiting role of HPT in the pathway, characterize the limitations in tocopherol biosynthesis in transplastomic leaves and fruits, and reveal a synthetic operon design that facilitates efficient multigene engineering in plastids.

Results

Expression of Three Enzymes of the Tocopherol Pathway from the Tobacco Plastid Genome.

Based on overexpression studies in nuclear-transgenic Arabidopsis plants, two enzymes in the tocopherol biosynthetic pathway (Fig. 1) were suggested as rate-limiting: homogentisate phytyltransferase [HPT, encoded by the VTE2 gene (5)] and tocopherol cyclase [TCY, encoded by the VTE1 locus (8)]. Recent data have called into question the possible limiting role of TCY in tocopherol biosynthesis (9, 10). Expression of foreign genes from the plastid genome has the advantages that (i) there is no interference from epigenetic processes, and (ii) transgene insertion occurs by homologous recombination into a preselected region of the plastid genome, typically an intergenic spacer region (11–13). Therefore, transgene expression is not burdened by copy number-dependent or insertion site-dependent variation in expression levels and epigenetic transgene silencing. However, because of lower plastid number per cell, lower plastid genome copy numbers per cell, and lower gene expression activity, transgene expression can be reduced significantly in nongreen tissues compared with photosynthetically active tissues.

We used plastid transformation in the model plant tobacco to test single-gene expression constructs for their effects on the flux through tocochromanol biosynthesis. To this end, we overexpressed the three key plastid-localized enzymes specific to tocopherol biosynthesis, HPT, TCY, and γ-tocopherol methyltransferase (TMT) (Fig. 1), in identical expression cassettes and targeted them to the same intergenic region in the plastid genome (Fig. 2). Both the expression cassette and the targeting site in the plastid DNA had been tested extensively in previous experiments and were shown to give highly efficient and stable transgene expression from the plastid genome (14–17). To avoid limitations from plant-specific regulation of enzyme activities, we used the genes encoding HPT, TCY, and TMT from the cyanobacterium Synechocystis sp. PCC6803 (18). For the final enzyme in the pathway, TMT [which is unlikely to limit flux through the tocochromanol pathway; (4)], we additionally used the gene from A. thaliana (Fig. 2A).

Fig. 2.

Introduction of constructs for metabolic engineering of the tocopherol pathway into the tobacco and tomato plastid genomes. (A) Physical maps of the transgene constructs and their targeting region in the tobacco plastid genome. Genes above the line are transcribed from the left to the right; genes below the line are transcribed from right to left. Relevant restriction sites and resultant fragments sizes in RFLP analyses of transplastomic lines are indicated. For gene names and encoded gene products, see Fig. 1. Two polycistronic constructs (pTop1 and pTop2) were designed by spacing the genes with either two intergenic regions from the plastid genome (IGR1 and IGR2 in pTop1) or two identical intercistronic expression elements (IEE in pTop2) (38). The transgene expression cassettes are driven by the ribosomal RNA operon promoter from tobacco (Nt-Prrn) fused to the 5′ UTR from phage T7 gene 10 (14,16) and the 3′ UTR of the plastid rbcL gene (Nt-TrbcL). The only exception is that in pTop2 the 3′ UTR of the Chlamydomonas reinhardtii plastid rbcL gene (Cr-TrbcL) was used as terminator of the operon, and the 3′ UTRs of the tobacco plastid rbcL and rps16 (Nt-Trps16) genes were used to stabilize the processed monocistronic mRNAs. The selectable marker gene aadA is driven by a chimeric ribosomal RNA operon promoter and is fused to the 3′ UTR from the plastid psbA gene (Nt-TpsbA) (19). Relevant restriction sites and fragment sizes are indicated. GOI: gene of interest; IGR1: intergenic region between psbH and petB; IGR2: intergenic region between rps2 and atpI. (B) Physical map of the targeting region in the wild-type tobacco plastid genome (N.t. ptDNA). Expected sizes of restriction fragments in RFLP analyses are indicated. (C) Physical map of the targeting region in the wild-type tomato plastid genome (S.l. ptDNA).

All four constructs were introduced into the tobacco plastid genome by particle gun-mediated transformation followed by selection for spectinomycin resistance conferred by a chimeric aadA gene present in all transformation vectors (Fig. 2A) (19). For each construct, two independently generated transplastomic lines were selected for in-depth analysis. The lines are referred to hereafter as “Nt-SyHPT” lines (for Nicotiana tabacum plants expressing the Synechocystis HPT enzyme, with Nt-SyHPT1 and Nt-SyHPT2 designating the two independently generated lines), “Nt-SyTCY” lines (expressing the Synechocystis TCY enzyme), “Nt-SyTMT” lines (expressing the Synechocystis TMT enzyme), and “Nt-AtTMT” lines (expressing the Arabidopsis TMT).

To test the transplastomic lines for homoplasmy (i.e., the absence of residual wild-type copies of the highly polyploid plastid genome), DNA samples were taken after two additional rounds of regeneration under antibiotic selection (20, 21) and analyzed by restriction fragment-length polymorphism (RFLP) analysis (Fig. S1). These experiments (i) confirmed successful transformation of the plastid genome in all tested lines, (ii) verified correct integration of the transgenes into the targeted genomic region by homologous recombination (Fig. 2 A and B), and (iii) provided preliminary evidence for homoplasmy of the transformed plastid genome (Fig. S1A). Homoplasmy ultimately was confirmed by seed assays (20) that yielded a uniform population of spectinomycin-resistant seedlings (Fig. S1C), consistent with maternal inheritance of the plastid genome in tobacco (22).

To verify expression of the transgenes from the plastid genome, a set of Northern blot experiments was conducted. Hybridizations with probes specific to the four transgenes yielded similar transcript patterns (Fig. 3A), in that two prominent hybridizing bands were detected in all transplastomic lines. The lower band corresponds to the expected size of the transgene transcript. The additional, slightly larger mRNA species originates from read-through transcription and has been observed previously in other studies using the same basic transformation vector pKP9 for transgene expression from the plastid genome (15, 16).

Fig. 3.

Analysis of mRNA accumulation in transplastomic tobacco lines expressing enzymes of the tocopherol biosynthetic pathway. (A) mRNA accumulation in transplastomic lines expressing monocistronic constructs. The transgene-specific hybridization probes detect two major transcript species in all lines. The smaller species of ∼1.5 kb corresponds to the monocistronic transcript from the transgene, whereas the larger transcript species originates from read-through transcription, as observed previously with similar transgenic constructs (15, 62). Sizes of the RNA marker bands are indicated in kilobases. (B) mRNAs accumulation in transplastomic lines that express the polycistronic constructs Top1 and Top2. Note that only in Nt-Top2 lines are the operon transcripts properly processed into three monocistronic mRNAs that can be detected readily with the individual hybridization probes. In contrast, Nt-Top1 lines accumulate a series of larger transcripts, including the unprocessed tricistronic precursor RNA of ∼3.8 kb.

When grown to maturity in soil under standard greenhouse conditions, none of the transplastomic lines displayed a discernible phenotype, indicating that chloroplast expression of the three enzymes of the tocopherol pathway has no obvious negative phenotypic consequences.

Tocochromanol Biosynthesis in Transplastomic Tobacco Plants Overexpressing the Three Pathway Enzymes from Single-Gene Constructs.

We next wanted to examine the effect of transgene expression in the transplastomic tobacco lines on tocochromanol metabolism. To this end, tocochromanols were extracted from leaf tissue with methanol and analyzed by reversed-phase HPLC using pure standards for quantitation (Fig. 4). Although no significant changes in tocochromanol accumulation were seen in the Nt-SyTCY, Nt-SyTMT, and Nt-AtTMT lines, a strong increase in tocopherol accumulation was observed in the Nt-SyHPT lines which express the Synechocystis HPT enzyme. Total tocochromanol levels were nearly fivefold higher in Nt-SyHPT plants than in wild-type plants (Fig. 4A), confirming previous data suggesting that HPT has a limiting role in tocopherol biosynthesis (5, 10). Most of the additional tocochromanol accumulating in the Nt-SyHPT lines was α-tocopherol, although some β-/γ-tocopherol (not normally accumulating in the wild type) was detectable also. This result indicates efficient conversion of γ-tocopherol to α-tocopherol in the HPT-expressing plants and argues against a serious limitation from TMT activity in leaf α-tocopherol synthesis. Accumulation of small amounts of tocotrienols (which are not detectable in the wild type) also was observed in the transplastomic HPT-expressing lines (Fig. 4A). This activity could be the result of some side activity of HPT as a geranylgeranyltransferase (Fig. 1), as observed previously in vitro for the Synechocystis HPT (23).

Fig. 4.

Accumulation of tocochromanol and photosynthetic pigment in transplastomic tobacco plants expressing enzymes of the tocopherol biosynthetic pathway. (A) Accumulation levels of the individual tocochromanols in leaves of wild-type plants (Nt-WT) and the various transplastomic lines generated in this study. The third leaf from the top of a mature plant (with a total of seven leaves) was harvested for the analysis. For each plant line, four different plants were measured. Error bars represent the SD. For each construct, two independently generated transplastomic lines were analyzed. Asterisks indicate values that are significantly different from the wild type (P < 0.05). Note that the β and γ forms (of both tocopherols and tocotrienols) could not be separated by chromatography and therefore are represented in a single bar. (B) Pigment contents in leaves of wild-type plants, transplastomic lines overexpressing HPT, and lines expressing the Top2 operon construct. Note that the levels of all pigments (carotenoids and chlorophylls) are slightly elevated in the transplastomic lines. Plants were grown under a light intensity of 500 µE⋅m⁻²⋅s⁻¹. (C) Cold stress recovery assay to compare the tolerance to oxidative stress in wild-type plants and transplastomic lines overaccumulating tocopherols (Nt-SyHPT and Nt-Top2). Plants were exposed to cold stress at 4 °C for 1 mo and then were transferred back to 25 °C for 1 wk. The first three leaves from the top were harvested and photographed. Although wild-type leaves show severe symptoms of photooxidative damage (evidenced by strong pigment loss starting from the leaf margins), the overproducing lines vitamin E recover without serious damage and show only slightly yellow leaf margins toward the tip.

Coexpression of Three Pathway Enzymes from Operon Constructs.

We next wanted to determine what limits tocopherol biosynthesis in our HPT-expressing plants. One possibility is that, in the Nt-SyHPT lines, HPT activity still limits tocopherol biosynthesis and that an even stronger HPT overexpression would be needed to enhance tocopherol synthesis further. Alternatively, HPT expression from the chloroplast genome may have removed the HPT bottleneck completely so that now another enzyme in the pathway becomes limiting. The most plausible candidate for such a secondarily limiting enzyme seems to be the tocopherol cyclase that acts downstream of HPT in the pathway (Fig. 1) (24). To test this possibility, we decided to combine all three chloroplast enzymes of the tocopherol biosynthetic pathway (HPT, TCY, and TMT) in a synthetic operon.

We initially constructed a bacterial-type operon, pTop1 (Fig. 2A) in which the expression of the operon genes is driven by a single promoter, the individual cistrons are separated by two intergenic regions taken from endogenous operons in the chloroplast genome (Materials and Methods), and the final cistron is followed by a 3′ UTR to stabilize the polycistronic mRNA (25). The operon construct was transformed into tobacco chloroplasts, and the resulting transplastomic lines (referred to as “Nt-Top1 lines”) were purified to homoplasmy (Fig. S1B). Analysis of RNA accumulation in Nt-Top1 transplastomic lines confirmed the presence of the expected tricistronic transcript of ∼3.8 kb but also revealed abundant shorter-than-expected RNA species (Fig. 3B). This observation suggests that some RNA cleavage and/or degradation of the polycistronic operon transcript occur.

HPLC analysis of leaf material from Nt-Top1 transplastomic lines was performed to determine how coexpression of the three tocopherol biosynthesis genes from the operon construct affects tocochromanol synthesis in tobacco. Surprisingly, expression of the tocopherol operon resulted in only a moderate increase (∼1.7-fold) in tocopherol accumulation. We suspected inefficient translation of the polycistronic mRNA to be the cause of the limited effect of the operon expression on the flux through tocopherol biosynthesis. Transcription of bacterial operons usually gives rise to stable polycistronic mRNAs which are translated directly. In contrast, many polycistronic precursor transcripts in plastids are posttranscriptionally processed into monocistronic or oligocistronic RNAs (26, 27), presumably by endonucleolytic cleavage (28). In at least some cases, cutting of the polycistronic precursor RNA into monocistronic units is required to facilitate efficient translation of all cistrons, as is supported by the analysis of nuclear mutants defective in distinct intercistronic processing events and by in vitro translation studies (29–32). Some polycistronic transcripts in plastids clearly remain unprocessed and are translated efficiently [e.g., the psbE operon comprising four small genes for polypeptides of photosystem II and the dicistronic psaA/B mRNA encoding the two reaction center subunits of photosystem I (33, 34)], but the rules determining the dependency of translation on intercistronic RNA processing are currently unknown. Thus which (trans)genes in synthetic operons will be expressed efficiently from unprocessed polycistronic transcripts and which transgenes will require intercistronic processing for efficient translation remains unpredictable, and previous attempts to stack transgenes in operons have produced mixed results (e.g., 35–37).

We recently have identified a small RNA element (dubbed the “intercistronic expression element,” IEE) that, when introduced into a chimeric context, proved to be sufficient to trigger processing of polycistronic transcripts into stable and translatable monocistronic mRNAs (38). Because we suspected the poor performance of our Top1 tocopherol operon in chloroplasts might be caused by inefficient translation of the polycistronic mRNA (which, in addition, was not fully stable; Fig. 3), we decided to redesign the operon and use the IEE element to enable cleavage of the operon transcript into stable monocistronic units. This redesigned synthetic operon, pTop2, was assembled from the coding regions of the three tocopherol genes, the promoter driving transcription of the operon, two copies of the 50-nt IEE element separating the three cistrons, and three different 3′ UTR sequences to confer stability of the processed mRNAs (Fig. 2A). Transplastomic Nt-Top2 plants were generated by chloroplast transformation in tobacco and purified to homoplasmy (as confirmed by Southern blot analysis and inheritance assays; Fig. S1 B and C). Transcript analyses by Northern blot experiments demonstrated that the IEE indeed triggered faithful processing of the polycistronic operon transcript into stable monocistronic mRNAs for all three cistrons (Fig. 3B), confirming that the IEE can serve as a sequence context-independent tool to facilitate efficient expression of synthetic operons from the plastid genome.

Having confirmed efficient processing and stable mRNA accumulation from the redesigned synthetic tocopherol operon, we next examined tocochromanol accumulation in the Nt-Top2 lines. Interestingly, the transplastomic Nt-Top2 plants showed an increase in tocochromanol accumulation that strongly exceeded the increase seen before in our Nt-SyHPT lines (Fig. 4A). The Nt-Top2 plants also showed a higher abundance of tocotrienols than the Nt-SyHPT lines (Fig. 4A). These data suggest that tocochromanol synthesis can be boosted further by coexpressing HPT with the downstream enzymes in the pathway and provide evidence that TCY expression limits tocopherol biosynthesis once the HPT bottleneck has been removed.

Pigment Synthesis and Stress Tolerance of Transplastomic Tobacco Plants with Enhanced Tocochromanol Biosynthesis.

Our expression studies of tocopherol biosynthetic enzymes in tobacco chloroplasts revealed two sets of transplastomic lines with strongly elevated tocochromanol accumulation: the Nt-SyHPT lines and the Nt-Top2 lines. Tocochromanol biosynthesis draws on precursors from the isoprenoid biosynthetic pathway (Fig. 1), which provides precursors to a variety of other pathways, including carotenoid, chlorophyll, phylloquinone, plastoquinone, and gibberellin biosyntheses (39). To test whether these pathways are affected by the strongly increased flux through the tocopherol pathway in Nt-SyHPT and Nt-Top2 plants, we measured photosynthetic pigment accumulation in the tocochromanol-overproducing transplastomic plants and tested the transplastomic lines in comparative growth assays under a variety of conditions.

Nt-SyHPT and Nt-Top2 plants grown under medium- to high-intensity light (300–1,000 µE⋅m⁻²⋅s⁻¹), displayed no discernible phenotype. Surprisingly, when photosynthetic pigment accumulation was analyzed, the transplastomic lines showed slightly higher levels of carotenoids and chlorophylls than seen in wild-type plants (Fig. 4B). This result suggests that no competition for isoprenoid substrates occurs in the tocopherol-overproducing plants, at least not to an extent that affects pigment biosynthesis by precursor limitation.

When plants were grown under unnaturally low light conditions (∼190 μE⋅m⁻²⋅s⁻¹), the transplastomic Nt-Top2 plants, but not the Nt-SyHPT plants, displayed a mild phenotype (Fig. S2). Young leaves in Nt-Top2 plants were light green, a phenotype that disappeared during development so that older leaves were indistinguishable from wild-type leaves (Fig. S2A). Consistent with the light-green phenotype under low-light conditions, the levels of all photosynthetic pigments were slightly lower in Nt-Top2 plants (but not in Nt-SyHPT plants) than in the wild type (Fig. S2B).

Previous work has implicated tocopherols in the protection of plants from oxidative stress (40–42). To test whether our tocopherol-overaccumulating Nt-SyHPT and Nt-Top2 transplastomic tobacco plants show increased tolerance to photooxidative stress, we performed cold-stress recovery assays under conditions known to promote greatly the production of reactive oxygen species in chloroplasts (43–45). Although wild-type plants showed severe symptoms of photooxidative damage (evidenced by leaf bleaching) under these conditions, both the Nt-SyHPT and the Nt-Top2 transplastomic plants were much less affected (Fig. 4C), suggesting that their high tocochromanol content confers strong protection from oxidative stress.

Expression of Tocopherol Biosynthetic Enzymes from the Plastid Genomes of Red-Fruited and Green-Fruited Tomato Varieties.

Tocochromanols of plant origin provide vitamin E to animals and humans and therefore represent essential components of the human diet. We previously reported the development of a plastid transformation protocol for tomato, a food crop with an edible fruit (46). To be able to compare the regulation of tocopherol biosynthesis and the rate-limiting steps in leaves versus fruits and, at the same time, to test if the results obtained in tobacco are transferable to an important food crop, we attempted to introduce the pSyHPT and pTop2 constructs that had led to greatly elevated tocopherol synthesis in transplastomic tobacco into the tomato plastid genome. (Because plastid transformation is considerably more laborious and time-consuming in tomato than in tobacco, initial construct testing and optimization was done in the tobacco system.)

Our previous work has shown that, during chloroplast-to-chromoplast conversion in tomato fruit ripening, plastid gene and transgene expression are strongly down-regulated at both the transcriptional and translational levels (47, 15). To avoid limitations from inefficient plastid gene expression and to distinguish between effects from tissue-specific regulation (fruit versus leaf) and effects from plastid differentiation (chloroplast versus chromoplast), we wanted to compare tocopherol metabolism in red-fruited and green-fruited tomato varieties expressing the pSyHPT and pTop2 constructs. Because the only established tomato plastid transformation protocol is for two closely related red-fruited tomato varieties bred in Brazil (46, 48), we first needed to develop protocols for green-fruited varieties (see Fig. S5). For two well-regenerating commercial varieties, “Dorothy’s Green” and “Green Pineapple”, we succeeded in obtaining transplastomic plants using optimized selection and regeneration protocols (49). Thus we were able to test whether green-fruited, chloroplast-containing tomato varieties express plastid transgenes more efficiently than red-fruited, chromoplast-containing varieties and to analyze tocochromanol biosynthesis comparatively in ripe green fruits and red fruits.

We introduced the two constructs that led to strongly elevated tocopherol synthesis in tobacco (the single-gene construct pSyHPT and the operon construct pTop2; Fig. 2) into the plastid genomes of the three tomato varieties that now can be transformed: the red-fruited variety IPA-6 and the green-fruited varieties Dorothy’s Green and Green Pineapple. The fruits of Green Pineapple differ from those of Dorothy’s Green in that they are smaller, somewhat darker green, and have less white (non–green plastid-containing) fruit flesh (see Fig. S5A).

Transplastomic tomato lines were generated with both constructs and for all three genotypes. The lines are designated Sl-IPA (for Solanum lycopersicum IPA-6), Sl-DG (for S. lycopersicum Dorothy’s Green), and Sl-GP (for S. lycopersicum Green Pineapple). Regenerated transplastomic plants were characterized by RFLP analyses (compare Fig. S3 A and B with Fig. 2), which verified correct integration into the targeted region of the tomato plastid genome (50) and provided preliminary evidence of homoplasmy. The homoplasmic state subsequently was confirmed by inheritance assays (Fig. S3C).

Northern blot analyses detected very similar transcript accumulation patterns in transplastomic tomato plants (Fig. S4), as seen previously in the corresponding transplastomic tobacco lines (Fig. 3). Importantly, the tricistronic operon transcripts in the Sl-IPA-Top2, Sl-DG-Top2, and Sl-GP-Top2 lines underwent correct processing into stable monocistronic mRNAs for all three genes of the operon, indicating that the IEE, originally identified in tobacco (38), also serves as a faithful RNA-processing element in other species.

Tocochromanol Biosynthesis in Leaves and Fruits of Transplastomic Tomato Plants.

Tocochromanol biosynthesis in the six sets of transplastomic tomato plants was analyzed in leaves and fruits (Fig. 5). To assess the dependence of tocopherol accumulation on leaf age and plant development as well, young leaves (the second or third leaf from the top) and mature (fully expanded but not yet senescent) leaves were analyzed comparatively. In line with the results obtained with HPT expression in transplastomic tobacco plants, the Sl-SyHPT lines showed a strong increase in leaf tocochromanol synthesis that was very similar in all three tomato genotypes (Fig. 5 A, C, and E). The increase in tocochromanol accumulation was more pronounced in mature leaves, where tocochromanol levels were approximately sixfold higher in the transplastomic lines than in the wild type. As in tobacco leaves, expression of the cyanobacterial HPT enzyme in transplastomic tomato plants induced the biosynthesis of tocotrienols, which normally do not accumulate in leaves of dicotyledonous plants (6). In all three tomato varieties, expression of the Top2 tocopherol operon led to an even stronger increase in leaf tocochromanol accumulation than seen with HPT expression alone, reaching levels up to tenfold higher than in the wild type (Fig. 5 A, C, and E).

Fig. 5.

Tocochromanol accumulation in leaves and fruits of red-fruited and green-fruited transplastomic tomato varieties expressing enzymes of the tocopherol biosynthetic pathway. For each cultivar, wild-type plants, transplastomic lines expressing HPT from a monocistronic construct, and transplastomic lines expressing the Top2 operon were analyzed. Two independently generated transplastomic lines per construct and four different plants per line were measured. Error bars represent the SD. Asterisks indicate values that are significantly different from the wild type (P < 0.05). Note that the β and γ forms of both tocopherols and tocotrienols could not be separated by chromatography and therefore are represented in a single bar. To determine leaf tocopherol contents, young (not yet fully expanded) leaves from the top of the plant and mature leaves from the bottom of the plant were comparatively analyzed. (A) Tocochromanol accumulation in leaves of the red-fruited variety IPA-6. (B) Tocochromanol accumulation in fruits of the red-fruited variety IPA-6. (C) Tocochromanol accumulation in leaves of the green-fruited variety Green Pineapple. (D) Tocochromanol accumulation in fruits of the green-fruited variety Green Pineapple. (E) Tocochromanol accumulation in leaves of the green-fruited variety Dorothy’s Green. (F) Tocochromanol accumulation in fruits of the green-fruited variety Dorothy’s Green. Note that both Dorothy’s Green and Green Pineapple are commercial green-fruited varieties that do not accumulate carotenoids but ripen normally (so-called “green when ripe” varieties) (63).

When tocochromanols were measured in ripe tomato fruits of the red-fruited variety IPA-6, the increase in tocochromanol accumulation in the transplastomic lines compared with the wild type was significantly lower than in leaf tissue (Fig. 5B). Also, no induction of tocotrienol biosynthesis was seen in red fruits. This result is consistent with a strong decline in plastid gene expression activity occurring upon chloroplast-to-chromoplast conversion, as has been observed previously (47, 15), but also could be the result of precursor limitation (i.e., lower GGDP availability in red fruits). Interestingly, ripe tomato fruits of the two green-fruited varieties Dorothy’s Green and Green Pineapple showed much higher tocochromanol accumulation than tomatoes of the red-fruited variety IPA-6 (Fig. 5 D and F). Total tocochromanol contents in fruits of HPT-expressing transplastomic Dorothy’s Green plants were twice as high as in transplastomic IPA-6 fruits, whereas transplastomic Green Pineapple fruits had threefold higher tocochromanol contents. These results correlate with Green Pineapple fruits having the highest chlorophyll content and thus presumably the highest activity of plastid gene expression (Fig. S5A). Unlike HPT expression in IPA-6, expression in Dorothy’s Green and Green Pineapple led to tocotrienol accumulation in fruits. This accumulation is consistent with higher HPT expression levels in green-fruited varieties enhancing the side activity of the Synechocystis HPT as a homogentisic acid geranylgeranyl transferase (HGGT) (Fig. 1) (23).

Surprisingly, coexpression of TCY and TMT with HPT from the Top2 operon did not result in a further increase in tocochromanol accumulation in tomato fruits (Fig. 5 B, D, and F). This result is in stark contrast to the situation in tobacco and tomato leaves, where expression of the tocopherol operon caused an additional boost in tocochromanol synthesis (Figs. 3A and 5 A, C, and E). This finding indicates that, despite its overexpression, HPT still represents the rate-limiting enzyme of the tocopherol pathway in the Sl-SyHPT and Sl-Top2 tomatoes and suggests that tocochromanol biosynthesis exhibits different bottlenecks in different tissues.

To follow the time course of tocochromanol accumulation during the fruit-ripening process, four different ripening stages were investigated in the red-fruited tomato variety IPA-6 and the green-fruited tomato variety Green Pineapple. Interestingly, although the tocochromanol levels increased continuously during the fruit-ripening process in both varieties (and in the wild-type as well as in Sl-SyHPT and Sl-Top2 tomatoes), the rate of increase was much higher in the transplastomic Green Pineapple fruits (Fig. S6). This result suggests that continued transgene expression in fruit chloroplasts largely accounts for the very high tocochromanol accumulation levels in ripe tomatoes of the green-fruited varieties (Fig. 5).

Discussion

Using tocochromanol (vitamin E) biosynthesis as an example, our work described here has uncovered efficient strategies by which metabolic pathway engineering can be performed by stable transformation of the plastid genome. Because vitamin E is an essential vitamin that has a diversity of beneficial physiological and disease-preventing properties (1), elevating the vitamin E content is an important goal of breeding and genetic engineering efforts (2, 3, 51–53). In addition to enhancing the nutritional value of crops, increased levels of vitamin E compounds can improve plant performance under stressful conditions that are linked to reactive oxygen species (Fig. 3C) (54). Our results confirm that the optimal approaches to maximize vitamin E contents differ in leafy tissues and nonleafy tissues. In leaves, expression of all three pathway enzymes from a synthetic operon proved to be superior to expression of the rate-limiting enzyme HPT alone (Figs. 3 and 5), perhaps because the efficient conversion of 2-methyl-6-phytylbenzoquinone and 2,3-dimethyl-5-phytyl-1,4-benzoquinone into tocopherols by TCY and TMT (Fig. 1) creates a metabolic sink that ultimately diverts more precursors into the pathway. In contrast, in both green and red tomato fruits, HPT expression triggered higher vitamin E synthesis than expression of the entire operon (Fig. 5). A possible reason is that HPT poses a stronger limitation to tocochromanol biosynthesis in fruits than in leaves, so that coexpression of the other pathway enzymes not only is unable to confer a further boost in vitamin E synthesis but even slightly reduces the beneficial effects of HPT expression. This effect could be caused by the two additional transgenes draining away plastid expression capacity that is known to be limited in tomato fruits (47, 15). It also is possible that precursor availability (e.g., phytyl phosphate from chlorophyll turnover; ref. 55) limits tocopherol biosynthesis more strongly in fruits than in leaves and in this way contributes to the different levels of vitamin E accumulation in different tissues (Fig. 5).

Previously, impressive results have been obtained by enhancing the tocopherol pathway via nuclear transformation. In addition, these studies provided interesting insights into limiting tocochromanol intermediates and rate-limiting biochemical reactions. For example, a comprehensive study in soybean seeds revealed that, at the metabolite level, the availability of PDP represents the most serious limitation (3). Our data obtained here suggest that, at least in green leaves, plastid engineering can be even more efficient than nuclear genetic engineering, especially with the synthetic operon design developed here. Nuclear expression of HPT in A. thaliana induced an up to 4.4-fold increase in leaf tocopherol (5), whereas in our plastid operon-expressing lines, an up to 10-fold increase was obtained (Fig. 5). However, the most significant advantages of plastid engineering lie in its unique precision (because of transgene integration by homologous recombination) and the possibility of stacking multiple transgenes in operons for coexpression as polycistronic mRNAs, which currently is not feasible in the nuclear genome. The functioning of the IEE (38) in very different sequence contexts, including nptII, yfp, and the three tocopherol biosynthesis genes tested here, suggests that the IEE triggers processing of polycistronic transcripts into stable and translatable monocistronic units in a sequence context-independent manner and thus provides a generally usable tool for the construction of synthetic operons that are expressed efficiently. Therefore, the synthetic operon design developed in this study will facilitate future synthetic biology applications in plastids, especially the construction of complex artificial operons that introduce novel biochemical pathways into plants. Because of the lack of epigenetic gene silencing and position effects in plastids, the transplastomic technology also entails an enormous reduction in the number of transgenic events that need to be generated and analyzed. For example, although 48 independent HGGT transgenic events had to be analyzed to identify a strong tocotrienol accumulation phenotype [in which tocotrienols accounted for 74% of the total content of tocochromanols of maize embryos (6)], generation of one or two transplastomic lines per construct usually suffices. The savings become even more significant when multiple transgenes need to be introduced, a process that in conventional nuclear transformation methodology typically involves cotransformation and analysis of even more events or the combination of individual transgenic events by crossing (3). Finally, because of maternal inheritance of the plastid DNA, plastid transformation also greatly reduces the risk of unwanted pollen transmission, thus improving the biosafety of transgenic plants. In this respect, we built an additional useful feature into our operon vector: two loxP sites that can be used to eliminate the aadA selectable marker gene by transiently crossing in a nuclear-encoded plastid-targeted Cre recombinase (56, 15).

A byproduct of this work has been the development of plastid transformation for green-fruited tomato varieties. The benefit of using green-fruited tomato varieties for the synthesis of nutraceuticals and pharmaceuticals is twofold. First, the long-term presence of chloroplasts in the fruit ensures a higher activity of plastid gene expression than in chromoplast-containing red fruits (47). Low levels of gene expression also were described for other nongreen plastid types (e.g., amyloplasts), although recent work has suggested strategies to overcome these limitations, at least partially (57–59). Second, green-fruited tomatoes can greatly simplify identity preservation (of transgenic tomatoes, nutritionally enhanced tomatoes, or “pharma” tomatoes), at least in countries where only red-fruited tomatoes are marketed. The expansion of the genotype range for tomato plastid transformation reported here represents an important step toward the application of the transplastomic technology in an important food crop on a more routine basis.

Analysis of carotenoid and chlorophyll accumulation in transplastomic plants overaccumulating vitamin E revealed an interesting regulatory connection between tocochromanol biosynthesis and the pathways of photosynthetic pigment biosynthesis. Surprisingly, both carotenoid and chlorophyll levels were found to be slightly increased in plants overproducing tocopherol, at least when grown under standard conditions (Fig. 3B). This observation may suggest that the increased flux through the tocochromanol pathway results in a general stimulation of isoprenoid-using pathways (possibly through positive feedback regulation). Alternatively, higher levels of protection from photooxidative stress could lead to reduced turnover of photosynthetic pigments. Interestingly, the plants expressing the synthetic tocopherol operon and showing the strongest overaccumulation of tocochromanols displayed a mild pigment deficiency under low-light conditions (Fig. S2). This deficiency indicates that the level of tocochromanol synthesis we have achieved with our transplastomic approach probably is close the maximum that is possible without negatively affecting metabolism and, hence, plant fitness. It is noteworthy in this respect that spinach, one of the vegetable plants with the highest vitamin E content, has approximately four- to fivefold higher tocochromanol levels than tomato fruits (60). Thus, the level in spinach is similar to the levels reached in our best-performing transplastomic tomatoes but is surpassed significantly by the levels reached in the leaves of transplastomic plants expressing the synthetic tocopherol operon.

In sum, our work presented here provides efficient strategies for enhancing the vitamin E content in leaves and fruits through plastid genome engineering and pinpoints principles of synthetic operon design for efficient metabolic pathway engineering and synthetic biology in plastids.

Materials and Methods

Plant Material and Growth Conditions.

Tobacco (N. tabacum cv. Petite Havana) plants were grown under aseptic conditions on agar-solidified medium containing 30 g/L sucrose (19). The tomato (S. lycopersicum) variety IPA-6 is a commercial red-fruited cultivar bred in Brazil. The green-fruited varieties Green Pineapple and Dorothy’s Green are presumably unrelated heirloom tomatoes (of unclear genetic origin) that were first described by tomato growers in the United States (http://t.tatianastomatobase.com:88/wiki/Green_Pineapple; http://t.tatianastomatobase.com:88/wiki/Dorothy%27s_Green). Both are commercial varieties that do not accumulate carotenoids but ripen normally (so-called “green when ripe” varieties). Tomato plants were raised from surface-sterilized seeds on the same medium with 20 g/L sucrose. Regenerated shoots from transplastomic lines were rooted and propagated on the same medium. Rooted homoplasmic plants were transferred to soil and grown to maturity under standard greenhouse conditions (relative humidity: 55%; day temperature: 25 °C; night temperature: 20 °C; diurnal cycle: 16 h light/8 h darkness; light intensity: 190–600 µE⋅m⁻²⋅s⁻¹). The homoplasmic state of the T1 generation was tested by germinating seeds from selfed transplastomic plants on medium with spectinomycin (500 mg/L for tobacco, 100 mg/L for tomato).

Construction of Plastid Transformation Vectors.

Monocistronic expression cassettes for enzymes of the tocopherol pathway were assembled in the plastid expression vector pHK20 (14). The operon constructs pTOP1 and pTOP2 were assembled from the genes cloned individually into the pHK20 backbone. The cloning procedures and the sources of genes and expression elements are described in detail in SI Materials and Methods. Synthetic oligonucleotides used are listed in Table S1. The sequences of the synthetic constructs were submitted to the GenBank Nucleotide Sequence Database at the National Center for Biotechnology Information and can be obtained under the accession nos. JX235342 (pSyHPT), JX235343 (pSyTCY), JX235344 (pSyTMT), JX235341 (pAtTMT), JX235345 (pTop1), and JX235346 (pTop2).

Plastid Transformation and Selection of Transplastomic Lines.

Young leaves from aseptically grown tobacco and tomato plants were bombarded with plasmid-coated 0.6-μm gold particles using a helium-driven biolistic gun (PDS1000He; BioRad) with the Hepta Adapter setup. Primary spectinomycin-resistant lines were selected on plant-regeneration medium containing 500 mg/L spectinomycin (19, 46, 48).

To develop a plastid transformation protocol for green-fruited tomato varieties, we tested a number of commercially grown green-fruited varieties for their tissue culture and regeneration properties. In these analyses, we identified two varieties (Dorothy’s Green and Green Pineapple), that displayed good regeneration rates from leaf explants and also showed acceptable rates of rooting on synthetic medium as well as good fruit and seed set in the greenhouse (Fig. S5). Regeneration medium and selection conditions for plastid transformation of green-fruited tomato varieties were as described previously for red-fruited varieties (46, 48). The plastid transformation efficiencies expressed as number of confirmed transplastomic lines (pSyHPT + pTop2) per number of selection plates (Petri dishes) with bombarded leaf pieces were five pSyHPT + three pTop2 events per 19 plates for IPA-6, nine pSyHPT + eight pTop2 events per 469 plates for Dorothy’s Green, and three pSyHPT + 10 pTop2 events per 674 plates for Green Pineapple.

Spontaneous spectinomycin-resistant tobacco and tomato lines were identified by double-selection tests on regeneration medium containing 500 mg/L spectinomycin and 500 mg/L streptomycin (19, 20) and subsequently were eliminated. Several independent transplastomic lines were generated for each construct and were subjected to one to three additional rounds of regeneration on spectinomycin-containing regeneration medium to enrich the transplastome and select for homoplasmic cell lines.

Isolation of Nucleic Acids and Hybridization Procedures.

Total plant DNA was extracted from fresh leaf samples by a cetyltrimethylammonium bromide-based method (15). For RFLP analysis by Southern blotting, 5-µg samples of total DNA were treated with restriction enzymes, separated in 1% agarose gels, and transferred onto Hybond XL nylon membranes (GE Healthcare) by capillary blotting. For hybridization, [α³²P]dCTP-labeled probes were produced by random priming (Multiprime DNA labeling kit; GE Healthcare).

Hybridizations were performed at 65 °C using standard protocols. Total plant RNA was isolated by a guanidine isothiocyanate/phenol-based method (peqGOLD TriFast; Peqlab). RNA samples (3 µg total RNA) were denatured, separated in denaturing formaldehyde-containing agarose gels (1%), and blotted onto Hybond XL nylon membranes (GE Healthcare) by capillary blotting.

A 550-bp PCR product generated by amplification of a portion of the psaB coding region using primers P7247 and P7244 (48) and a 283-bp PCR product generated by amplification of the psbZ coding region using Pycf9a and Pycf9b (38) were used as RFLP probes to verify plastid transformation and assess homoplasmy. Transgene-specific probes for Northern blot analyses were generated by excising the complete coding regions from plasmid clones.

HPLC Analysis of Pigments and Tocochromanols.

Tocochromanols were extracted with methanol from lyophilized leaf powder or lyophilized tomato fruit powder. Separation, identification, and quantification of tocochromanols were carried out with an YMC ODS-A 250 × 4.6 mm column using the Agilent 1100 Series HPLC system. Separation was performed according to published procedures (42). Reversed-phase HPLC was used to avoid coelution of γ-tocopherol with plastochromanol-8 (9). Carotenoids and chlorophylls were isolated from lyophilized leaf powder samples by extraction with 80% (vol/vol) acetone followed by two extractions with 100% acetone. The three extracts subsequently were combined and used for HPLC analysis. Separation, identification, and quantification of pigments were performed as described previously (61) using the HPLC system specified above. All tocochromanol and pigment species were identified and quantified by comparison with known amounts of pure standards.

Cold Stress and Recovery Assay.

To assess and compare the tolerance of transplastomic and wild-type plants to oxidative stress, greenhouse-grown plants were transferred to a growth chamber adjusted to cold stress conditions (temperature: 4 °C; day length: 16 h; light intensity: 70 µE⋅m⁻²⋅s⁻¹). Following exposure to cold stress for 1 mo, plants were transferred back to normal growth conditions (temperature: 25 °C; light intensity: 250 μE⋅m⁻²⋅s⁻¹) for 1 wk. The top three leaves then were phenotypically compared and photographed.