Contained metabolic engineering in tomatoes by expression of carotenoid biosynthesis genes from the plastid genome

Authors


(fax +49 331 567 8701; e-mail rbock@mpimp-golm.mpg.de).

Summary

Applications of chloroplast engineering in agriculture and biotechnology will depend critically on success in extending the crop range of chloroplast transformation, and on the feasibility of expressing transgenes in edible organs (such as tubers and fruits), which often are not green and thus are much less active in chloroplast gene expression. We have improved a recently developed chloroplast-transformation system for tomato plants and applied it to engineering one of the central metabolic pathways in fruits: carotenoid biosynthesis. We report that plastid expression of a bacterial lycopene β-cyclase gene results in herbicide resistance and triggers conversion of lycopene, the main storage carotenoid of tomatoes, to β-carotene, resulting in fourfold enhanced pro-vitamin A content of the fruits. Our results demonstrate the feasibility of engineering nutritionally important biochemical pathways in non-green plastids by transformation of the chloroplast genome.

Introduction

Chloroplast genetic engineering offers several attractions to plant biotechnologists, including high-level foreign protein accumulation, absence of epigenetic effects, convenient transgene stacking in operons, and gene containment due to the lack of transgene transmission through pollen (Bock, 2001; Daniell et al., 2002; Maliga, 2004). The procedures are technically demanding and, despite recent progress in establishing protocols for some food crops (Dufourmantel et al., 2004; Kumar et al., 2004; Lelivelt et al., 2005; Ruf et al., 2001; Sidorov et al., 1999), plastid transformation is routine only in the model system tobacco (Maliga, 2004). The implementation of plastid genome engineering in agriculture and biotechnology will depend critically on (i) methodological improvements that make the technology routine in important food crops, and (ii) success with the expression of chloroplast transgenes in edible plant organs such as tubers and fruits (Bock and Khan, 2004; Maliga, 2003). The latter is of pivotal importance because most endogenous plastid genome-encoded genes are involved in photosynthesis and hence are drastically downregulated in all non-photosynthetic tissues.

Plant metabolic engineering has great potential for the production of safer and more nutritious foods. In view of the attractions of plants with transgenic chloroplasts (transplastomic plants), the engineering of nutritionally important metabolic pathways via plastid transformation seems a very worthwhile challenge. Having recently developed a transplastomic technology for tomato plants (Ruf et al., 2001), we have begun to engineer by plastid transformation one of the key metabolic pathways in tomatoes: carotenoid biosynthesis.

Carotenoids are isoprenoid molecules that are very widespread in nature. They are typically seen as colourful pigments in fruits and flowers of plants, but also occur in animals such as crustaceans and birds. Carotenoids are essential components of the photosynthetic membranes in plants and algae, although they are usually not seen in green tissues due to masking by the chlorophylls. Animals are unable to synthesize carotenoids de novo, and therefore must rely on their diet as a source of these compounds. The carotenoid β-carotene is an important antioxidant and the major dietary precursor of vitamin A. The US recommended dietary allowance for vitamin A is 700–900 μg day−1 for adults (http://www.nlm.nih.gov/medlineplus/druginfo/natural/patient-vitamina.html). Vitamin A deficiency is a major public health problem in about one-third of the countries of the world. It causes severe symptoms, including night blindness, xerophthalmia, increased susceptibility to respiratory, gastrointestinal and childhood diseases and, if untreated, leads to complete blindness (Ye et al., 2000). In addition, therapeutic doses of β-carotene can have protective effects against cardiovascular diseases, certain cancers and ageing-related diseases (Collins, 1999; Krinsky, 1998). Therefore increasing the pro-vitamin A content of food crops represents an important goal of breeding and genetic engineering efforts (Giuliano et al., 2000; Hirschberg, 1999; Paine et al., 2005; Römer et al., 2000; Ye et al., 2000). In this work, we have attempted to elevate the pro-vitamin A content of tomatoes by genetic engineering of the chloroplast genome.

Results

Construction of new vectors and optimization of tomato chloroplast transformation

Our previously described method for tomato chloroplast transformation was relatively laborious and time-consuming, and yielded transplastomic lines at much lower frequency than the well-established tobacco system (Ruf et al., 2001). We have therefore sought to develop an improved protocol that would make tomato plastid transformation less laborious and allow the use of the technology on a routine basis. One of the technical difficulties with tomato chloroplast transformation comes from the need to bombard large surface areas of leaf material, making it impractical to use highly regenerable cotyledons, which are the standard source material for nuclear transformation experiments in tomato using Agrobacterium tumefaciens. Improving the efficiency of plant regeneration from true leaves was therefore key to the optimization of tomato plastid-transformation procedures. In order to identify tomato genotypes that display higher regeneration rates from leaves than the previously used cv. Santa Clara, we have tested a variety of tomato cultivars for their shoot-induction response to different tissue culture media. In the course of this work, we have identified a new combination of tomato genotype (cv. IPA-6) and regeneration medium (RMOL) that yields greatly enhanced plant regeneration from true leaves (see Experimental procedures). When tested in chloroplast-transformation experiments, the improved regeneration protocol yielded transplastomic tomato lines at high frequency (on average one to two transformants per shot with the PDS1000He/Hepta adaptor biolistic device), and had the additional advantage over the previous protocol that direct shoot regeneration was obtained for >50% of the transplastomic clones already in the primary selection.

Our previously constructed series of plastid-transformation vectors (pRB94 and pRB95; Ruf et al., 2001) harboured the aadA selectable marker gene and a polylinker with unique restriction sites. To facilitate the direct insertion of passenger genes, we have now modified these vectors by insertion of an expression cassette consisting of the plastid atpI promoter and the 3′ UTR from the rps16 gene (Figure 1). The atpI promoter was selected to facilitate transgene expression in tomato fruits, which contain carotenoid-accumulating chromoplasts instead of photosynthetically active chloroplasts. The atpI promoter contains multiple binding sites for both RNA polymerases present in higher plant plastids, and one of the mapped transcription-initiation sites has been shown also to be active in non-photosynthetic tissues (Miyagi et al., 1998). The coding region of any gene of interest can be integrated into pRB96 in a single cloning step as an NcoI/XbaI fragment, with the NcoI restriction site containing the start codon and the XbaI site containing the stop codon of the reading frame.

Figure 1.

 Physical map of the targeting region in the plastid genome and structure of chloroplast transformation vectors.
(a) Map of the plastid-targeting region from which the chloroplast transformation vectors were derived. The transgenes are targeted to the intergenic region between the trnfM and trnG genes.
(b) Map of the basic chloroplast transformation vector pRB96. Plasmid pRB96 is a derivative of pRB94 (Ruf et al., 2001) and differs from the latter in that it harbours a transgene expression cassette within the polylinker. The cassette consists of the promoter and 5′ UTR from the plastid atpI gene and the 3′ UTR of the rps16 gene. The transgenes integrated as genes of interest in the course of this study are indicated above the vector map and abbreviated as follows: PSP, phytoene synthase gene from Phycomyces blakesleeanus; LCP, lycopene β-cyclase gene from P. blakesleeanus; LPP, lycopene β-cyclase/phytoene synthase fusion gene from P. blakesleeanus; LCE, lycopene β-cyclase from Erwinia herbicola. All transgenes were engineered to contain a 5′NcoI and a 3′XbaI restriction site, and integrated as NcoI/XbaI fragments into the PatpI/Trps16 expression cassette. Relevant restriction sites are marked. Sites lost due to ligation to heterologous ends are in parentheses .
(c) Carotenoid biosynthetic pathway in plants.

Introduction of carotenoid biosynthesis genes into the tobacco and tomato plastid genomes

Microbial carotenoid biosynthesis genes have proven to be a good source material for engineering the carotenoid biosynthetic pathway in plants, the generation of ‘Golden Rice’ being a prime example (Al-Babili and Beyer, 2005; Ye et al., 2000). Lycopene is the major storage carotenoid of the tomato fruit, and can be converted into pro-vitamin A (β-carotene) by two cyclization reactions that create one β-ring at each end of the lycopene molecule (Figure 1c). This reaction is catalysed by lycopene β-cyclase, an enzyme that is not present in animals (including humans), which therefore cannot convert lycopene to β-carotene and are dependent on dietary uptake of pro-vitamin A.

In view of the nutritional importance of vitamin A and the high amounts of lycopene present in tomatoes, we selected two microbial lycopene β-cyclase genes for engineering carotenoid biosynthesis in tomato chromoplasts. In addition to the lycopene β-cyclase gene (crtY) from the carotenoid-producing eubacterium Erwinia herbicola, we selected the carRA gene from the zygomycete fungus Phycomyces blakesleeanus. This gene is particularly interesting in that the encoded protein has been suggested to be bifunctional (Arrach et al., 2001): its N-terminal domain carries lycopene β-cyclase activity, whereas the C-terminal domain harbours phytoene synthase activity. The protein has been suggested to undergo post-translational processing by proteolytic cleavage, separating the two enzyme activities into distinct polypeptide chains (Arrach et al., 2001).

Using the Erwinia crtY and Phycomyces carRA genes, we constructed four vectors for chloroplast transformation (Figure 1b). In addition to crtY and the full-length carRA, we also designed single-enzyme constructs for the Phycomyces carRA gene by separating the lycopene β-cyclase and phytoene synthase domains at the proposed protein-cleavage site (Arrach et al., 2001) and inserting the corresponding cDNA sequences into plastid-transformation vector pRB96 (Figure 1).

All four constructs were introduced into both tobacco and tomato by biolistic plastid transformation (Ruf et al., 2001; Svab and Maliga, 1993). Several transplastomic lines were obtained for each construct, and two independently generated lines per construct and species were characterized in detail. The lines are referred to subsequently as PSP (lines carrying the phytoene synthase gene from P. blakesleeanus), LCP (lines carrying the lycopene β-cyclase gene from Phycomyces), LPP (lines carrying the lycopene β-cyclase/phytoene synthase fusion gene from Phycomyces), and LCE (lines carrying the lycopene β-cyclase from E. herbicola) with the prefix Nt indicating tobacco (Nicotiana tabacum) transplastomic lines and Sl designating tomato (Solanum lycopersicum) lines.

Analysis of transplastomic tobacco and tomato plants harbouring microbial carotenoid biosynthesis transgenes

Successful chloroplast transformation was verified by tests for double resistance on medium containing two aminoglycoside antibiotics, spectinomycin and streptomycin (Bock, 2001; Svab and Maliga, 1993) and further confirmed by PCR assays using transgene-specific primers (data not shown). Chloroplast transformants were subsequently purified to homoplasmy by passing them through additional regeneration cycles under antibiotic selection. Homoplasmy of the transplastomic lines (absence of any residual copies of the wild-type chloroplast genome) was assessed by RFLP analyses (Figure 2).

Figure 2.

 RFLP analysis of tobacco and tomato chloroplast transformants.
Total cellular DNA was digested with BamHI and hybridized to a radiolabelled probe detecting the region of the plastid genome that flanks the transgene-insertion site.
(a) Southern blot analysis of tobacco transplastomic lines carrying carotenoid biosynthesis genes. Fragment sizes for the wild type and the different transplastomic lines are indicated in kb.
(b) Southern blot analysis of tomato transplastomic lines carrying carotenoid biosynthesis genes from Phycomyces.
(c) Southern blot analysis of tomato transplastomic lines carrying the lycopene β-cyclase gene from Erwinia. A faint wild-type-like band is seen in all transgenic lines in (a, b): this band comes from promiscuous DNA in the nuclear genome (see text for details). The band is not detected when a short PCR product derived from the psaB coding region is used as a probe (c; see Experimental procedures), presumably because the promiscuous DNA is not sufficiently homologous to this region.

In addition to a strong band for the transplastomic fragment, RFLP analysis of transplastomic tobacco and tomato lines also yielded a faint hybridization signal that corresponds in size to the restriction fragment from the wild-type genome (Figure 2a,b). Persistence of a wild-type-like hybridization signal, even after multiple rounds of selection, is often seen in transplastomic lines and usually is caused not by true heteroplasmy of the chloroplast transformants, but rather by the presence of promiscuous plastid DNA in one of the other two genomes of the plant cell. It is well established that, during evolution, large fragments of chloroplast DNA have integrated into the nuclear and mitochondrial genomes (for reviews see Bock, 2005; Leister, 2005; Timmis et al., 2004) and this non-functional, so-called promiscuous DNA can produce wild-type-like bands in DNA gel-blot analyses of otherwise homoplasmic transplastomic lines (Hager et al., 1999; Ruf et al., 2000).

To ultimately confirm homoplasmy of all transplastomic lines, we performed seed assays, the most sensitive tests available to assess homoplasmy (Bock, 2001; Maliga, 2004). Lack of segregation of antibiotic resistance in the T1 generation demonstrated homoplasmy (Figure 3a,b) and confirmed uniparental maternal transgene inheritance, as expected for a plastid trait in tobacco and tomato (Hagemann, 2002; Maliga, 2004).

Figure 3.

 Seed assays to confirm homoplasmy of transplastomic lines and phenotypic analysis of transplastomic tobacco plants.
(a, b) Analysis of the T1 generation in tobacco (a) and tomato (b) transplastomic lines. All crosses with a transplastomic line as maternal parent yield progeny that is homogeneously resistant to spectinomycin, demonstrating homoplasmy of the transplastomic lines and confirming uniparental maternal inheritance of plastids in tobacco and tomato.
(c) Wild-type-like phenotype of homoplasmic transplastomic tobacco lines expressing carotenoid biosynthesis transgenes.

We next investigated whether chloroplast expression of microbial carotenoid biosynthesis genes results in altered phenotypes of transplastomic plants. When grown under a variety of different light regimes, all transplastomic tobacco and tomato lines were indistinguishable from the respective wild-type controls (Figure 3c; data not shown), indicating that transgene expression is phenotypically neutral.

To test for expression of the introduced carotenoid biosynthesis genes, we performed a series of Northern blot experiments (Figure 4) using transgene-specific probes. Hybridization to a P. blakesleeanus lycopene β-cyclase probe revealed a complex pattern of hybridizing bands, with multiple shorter-than-normal RNA species, for both the transplastomic lines with the lycopene β-cyclase gene and those with the lycopene β-cyclase/phytoene synthase fusion gene from Phycomyces (Figure 4a). This is indicative of massive RNA degradation, presumably due to instability of the transcript from the Phycomyces lycopene β-cyclase gene in chloroplasts. By contrast, hybridization to the Phycomyces phytoene synthase and the Erwinia lycopene β-cyclase probes yielded two highly abundant transcript species, the shorter one corresponding to the expected size of the full-length transcript, and the longer one corresponding to the size of the co-transcript with the downstream aadA marker gene (Figures 1b and 4b,c). This co-transcript originates from read-through transcription, which is commonly observed in transplastomic lines that carry selectable marker genes and genes of interest in the same orientation (Zoubenko et al., 1994). The relatively high abundance of the co-transcript may be due to the rps16 3′ UTR acting as a rather weak transcriptional terminator (Staub and Maliga, 1994).

Figure 4.

 Analysis of RNA accumulation in tobacco (Nt lines; left panels) and tomato (Sl lines; right panels) chloroplast transformants.
Total cellular RNA was hybridized to radiolabelled probes corresponding to the coding regions of the introduced carotenoid biosynthesis transgenes.
(a–c) Hybridization to (a) Phycomyces lycopene β-cyclase probe; (b) Phycomyces phytoene synthase probe; (c) Erwinia lycopene β-cyclase probe. The presence of a second strongly hybridizing band in (b, c) is probably due to read-through transcription generating a dicistronic transcript with the aadA as second cistron (see text). The transcript pattern in (a) is more complex, and shows evidence of transcript instability. Transcripts from the fungal lycopene β-cyclase/phytoene synthase fusion gene seem to be particularly unstable, with little full-length mRNA accumulating in (a) and very little mRNA being detectable in (b). M, molecular weight marker (kb).

Transcripts from the Phycomyces lycopene β-cyclase/phytoene synthase fusion gene are barely detectable with the phytoene synthase probe, and the absence of aberrant shorter-than-normal RNA species strongly suggests that transcript instability comes predominantly from the lycopene β-cyclase-encoding part of the mRNA (Figure 4a,b).

Carotenoid biosynthesis in transplastomic tobacco and tomato plants

To test whether expression of the microbial carotenoid biosynthesis genes led to altered carotenoid contents or spectra in transplastomic tobacco and tomato plants, we first tested for the presence of elevated lycopene β-cyclase activity in transplastomic plants. To this end, we challenged plants with 2-(4-chlorophenylthio)-triethylamine (CPTA), a herbicide that specifically inhibits lycopene β-cyclase activity (Schuetz and Baldwin, 1958; Tao et al., 2004). Just like the endogenous plant enzyme, the Erwinia lycopene β-cyclase is also sensitive to CPTA, so that increased levels of herbicide tolerance should correlate directly with increased amounts of the enzyme in the transplastomic plants.

Tolerance tests with wild-type tobacco revealed that watering of young plants with as little as 10 ml 50–100-μm CPTA solution was sufficient to cause severe chlorosis and ultimate death of the plants (Figure 5a). The herbicide predominantly affected the young growing leaves (Figure 5), which is readily explained by the essential function of β-carotene for the assembly of the photosynthetic apparatus during leaf development. When the transplastomic plants expressing carotenoid biosynthesis transgenes from Phycomyces were assayed for their tolerance to CPTA, no elevated tolerance could be observed (Figure 5a). Given the observed instability of the Phycomyces lycopene β-cyclase transcripts in chloroplasts, these results are unsurprising. In contrast, expression of the Erwinia lycopene β-cyclase resulted in significantly enhanced herbicide tolerance. Transplastomic tobacco plants treated with 50 μm CPTA showed no symptoms of herbicide susceptibility; plants exposed to 100 μm CPTA, although developing mild symptoms, also survived the herbicide treatment (Figure 5; data not shown). CPTA tolerance through expression of the Erwinia lycopene β-cyclase was also confirmed for transplastomic tomato plants (Figure 5b). These data suggest that expression of the bacterial lycopene β-cyclase indeed results in the presence of elevated enzyme levels in chloroplasts of transplastomic plants. Assuming that the endogenous plant enzyme and the Erwinia enzyme are about equally sensitive to CPTA, we cautiously estimate from comparative herbicide-tolerance assays with different CPTA concentrations (Figure 5; data not shown) that we have achieved an approximately fivefold overexpression of the lycopene β-cyclase in our transplastomic lines.

Figure 5.

 Herbicide-resistance assays to test for expression of lycopene β-cyclase transgenes at the protein level. The herbicide CPTA is a specific inhibitor of lycopene β-cyclase activity.
(a) Herbicide-tolerance tests in tobacco. Plants watered once with 10 ml CPTA solution (50 or 100 μm) were compared after 0, 5 and 10 days with a water-treated control (0 μm CPTA). Plastid expression of the Erwinia lycopene β-cyclase leads to strongly enhanced CPTA tolerance, with plants treated with 50 μm herbicide showing no symptoms after 10 days, and plants treated with 100 μm herbicide displaying only mild symptoms.
(b) CPTA tolerance in tomato by plastid expression of the Erwinia lycopene β-cyclase. Plants watered twice with 100 ml CPTA solution (100 μm) were compared with a water-treated control (0 μm CPTA) and photographed 10 days after the first herbicide application.

We next wanted to test whether the presence of increased amounts of enzyme correlates with an increase in lycopene β-cyclase activity in transplastomic plants. To this end, we determined pigment contents in transplastomic lines expressing the Erwinia lycopene β-cyclase and compared them with the pigment contents of wild-type plants and the transplastomic plants with the Phycomyces carotenoid biosynthesis genes. When tobacco and tomato leaf tissue was analysed, no significant differences were seen in either total carotenoid contents or abundance of individual carotenoid species (Tables 1 and 2; data not shown). This was expected, because no significant amounts of lycopene accumulate in leaves and therefore the presence of excess lycopene β-cyclase is unlikely to induce changes in the carotenoid spectrum. However, as lycopene represents the main storage carotenoid in tomato fruits, the presence of elevated levels of lycopene β-cyclase activity should increase conversion of the dark red carotenoid lycopene to the orange pigment β-carotene (pro-vitamin A). Consequently, if high amounts of lycopene were converted to β-carotene, the altered carotenoid composition should be phenotypically visible as a change in tomato fruit colour. Indeed, when transplastomic tomatoes from plants expressing the bacterial lycopene β-cyclase were compared with fruits from wild-type plants, both pericarp and fruit flesh were significantly more orange (Figure 6), indicating that increased lycopene to β-carotene conversion does occur. This overaccumulation of β-carotene could largely be blocked by treatment of fruits with CPTA concentrations sufficiently high to block both endogenous and transgene-derived lycopene β-cyclase activities (and resulted in a reversion of fruit colour to deeper red; see Experimental procedures and Figure 6), confirming that the elevated levels of β-carotene are due to transplastomic expression of the Erwinia lycopene β-cyclase.

Table 1.   Carotenoid contents (μg g−1 DW) in tobacco leaves from wild-type (WT) and transplastomic plants
 WTPSPLCPLPPLCE
  1. Values are means from four independent pigment extractions and HPLC analyses ±SD.

Neoxanthin98.1 ± 21.7111.2 ± 33.8123.5 ± 27.7104.1 ± 25.792.8 ± 23.7
Violaxanthin244.1 ± 39.7252.1 ± 22.2260.4 ± 23.2228.9 ± 35.2211.5 ± 58.5
Antheraxanthin12.7 ± 4.510.3 ± 3.513.1 ± 2.111.1 ± 1.110.8 ± 2.5
Lutein600.9 ± 72.8647.6 ± 121.2700.9 ± 95.9585.4 ± 94.8557.7 ± 103.1
Chlorophyll b2400.1 ± 321.62361.2 ± 361.82497.1 ± 440.62155.1 ± 311.82188.7 ± 367.7
Chlorophyll a6110.6 ± 884.16160.1 ± 901.96528.6 ± 1028.25558.4 ± 902.15443.9 ± 902.1
β-Carotene845.7 ± 187.5940.1 ± 198.41035.5 ± 189.5806.8 ± 181.1764.4 ± 173.8
Total carotenoid1801.6 ± 282.91961.6 ± 373.32133.6 ± 322.31736.4 ± 328.21637.4 ± 317.4
Total chlorophyll8510.7 ± 1205.78521.3 ± 12639025.7 ± 1465.27713.5 ± 1213.97632.6 ± 1267.8
Table 2.   Comparison of lycopene and β-carotene contents (μg g−1 DW) in tomato leaves and fruits from wild-type (WT) and transplastomic plants expressing the lycopene β-cyclase gene from Erwinia (LCE)
 LeafFruit
WTLCEWTLCE
  1. Values for leaves are means from four independent pigment extractions and HPLC analyses; values for fruits are means from three independent carotenoid extraction experiments with two HPLC measurements each, ±SD. nd, Not detectable.

Neoxanthin97.3 ± 6.686.4 ± 13.2ndnd
Violaxanthin370.8 ± 8.3340.8 ± 39.3ndnd
Antheraxanthin166.3 ± 44.1161.3 ± 34.6ndnd
Zeaxanthin81.4 ± 27.687.5 ± 25ndnd
Chlorophyll b1982.9 ± 2811999.5 ± 175.1ndnd
Chlorophyll a4737.0 ± 549.54427.9 ± 431.8ndnd
Lycopenendnd3657.4 ± 156.52951 ± 138.4
β-Carotene1057.2 ± 85915.1 ± 161.269.1 ± 21.4286.1 ± 14.6
Total carotenoid1773.1 ± 171.61591.1 ± 242.73726.6 ± 159.73237.1 ± 151.6
Total chlorophyll6719.8 ± 830.46427.4 ± 160.9ndnd
Figure 6.

 Phenotype of ripe tomato fruits harvested from transplastomic tomato plants expressing the lycopene β-cyclase gene from Erwinia.
Upper row, fruits from the wild type; middle row, fruits from transplastomic plants; bottom row, fruits from transplastomic plants treated with 2.5 mm CPTA. The more orange colour of both pericarp and fruit flesh in the transplastomic tomatoes suggests conversion of a significant amount of red lycopene into orange β-carotene (pro-vitamin A). This conversion and the associated colour change can be largely blocked by treatment with CPTA.

HPLC analyses of carotenoid accumulation in fruits revealed that tomatoes from the transplastomic lines expressing the Erwinia lycopene β-cyclase had, on average, a fourfold enhanced pro-vitamin A content (Table 2). As expected, this increase in β-carotene accumulation occurred at the expense of lycopene accumulation, which was found to be reduced by approximately 10–15% (Table 2).

To confirm that the plastid Erwinia lycopene β-cyclase gene is still actively expressed in fruits, we analysed a developmental series of tomato fruits ranging from very small green fruits to fully ripe red tomatoes (Figure 7). RNA abundance in fruits is lower than in the leaf, consistent with our previous observation that foreign protein accumulation in tomato fruit chromoplasts is somewhat lower than in chloroplasts (Ruf et al., 2001). Interestingly, while transcript levels decline from early fruit development to the beginning of chloroplast-to-chromoplast transition, they remain about constant from the breaker stage onwards (Figure 7), suggesting that the major adjustments in plastid gene expression during fruit development may occur before the onset of ripening.

Figure 7.

 Time course of lycopene β-cyclase expression in tomato fruits.
Total cellular RNA was extracted from leaves and various stages of fruit development and hybridized to a radiolabelled probe corresponding to the coding region of the Erwinia lycopene β-cyclase. M, molecular weight marker (kb).

In contrast, the carotenoid contents of tomatoes from the transplastomic plants expressing the Phycomyces carotenoid biosynthesis genes were unaltered (data not shown), which is in line with the results from our herbicide-tolerance assays, the observed transcript instability of the Phycomyces lycopene β-cyclase in plastids (Figure 4), and the unchanged fruit colour of ripe tomato fruits (data not shown).

Discussion

In this work, we have tested four carotenoid biosynthesis genes as plastid transgenes in both tobacco and tomato plants. While the three fungal gene constructs did not result in measurable changes of carotenoid biosynthesis in leaves and fruits, expression of a bacterial lycopene β-cyclase led to a fourfold enhanced pro-vitamin A content in transplastomic tomatoes.

The unsuccessful expression of the Phycomyces lycopene β-cyclase constructs is probably attributable to the observed instability of the corresponding transcripts in plastids (Figure 4). Although the reason for this instability remains unclear, the presence of cryptic RNA-processing sites within the coding region, which would induce aberrant RNA cleavage, could provide a possible explanation and would be compatible with the observed presence of numerous shorter-than-expected RNA species (Figure 4). Whether or not fungal genes (or cDNAs) are, in general, more problematic to express in plastids than, for example, bacterial genes remains to be tested. However, the recent successful expression of a yeast trehalose phosphate synthase in tobacco chloroplasts (Lee et al., 2003) indicates that mRNA instability is at least not a universal phenomenon associated with the expression of fungal genes from the plastid genome.

In contrast to the two constructs harbouring lycopene β-cyclase sequences, transcripts from the Phycomyces phytoene synthase were stable and accumulated to high levels in both tobacco and tomato plastids (Figure 4). Nonetheless, our carotenoid analyses provided no evidence for increased phytoene synthase activity in transgenic plastids (Table 1; data not shown). Earlier work using conventional nuclear transformation has shown that overexpression of phytoene synthase can trigger increased carotenoid synthesis (Fraser et al., 2002). The obvious absence of elevated phytoene synthase activity in our transplastomic lines could be due to lack of translation of the mRNA, protein instability in plastids, or enzymatic inactivity of the protein. The latter possibility is conceivable because the phytoene synthase activity is part of a bifunctional protein in Phycomyces, and the enzymatic activity may depend on the full-length protein (although post-translational processing into two distinct polypeptide chains has been postulated; Arrach et al., 2001).

Our finding that expression of the lycopene β-cyclase gene from the eubacterium Erwinia led to a fourfold enhanced pro-vitamin A content in transplastomic tomatoes demonstrates the feasibility of engineering nutritionally important metabolic pathways in non-green plastids by transformation of the chloroplast genome. Most endogenous plastid genome-encoded genes are involved in photosynthesis, and hence are strongly downregulated in non-photosynthetic tissues. Ripe red tomato fruits contain chromoplasts which, during fruit ripening, develop from the chloroplasts present in green tomatoes. Differentiation of chromoplasts from chloroplasts, which are highly active in gene expression, may be one of the factors responsible for plastid transgene expression in tomatoes being more successful than, for example, in potatoes (Ruf et al., 2001; Sidorov et al., 1999).

All previous work on engineering the carotenoid biosynthesis pathway in plants was conducted by expressing transgenes from the nuclear genome. Two strategies have been pursued: introduction of bacterial genes for enzymes of the pathway (Fraser et al., 2002; Götz et al., 2002; Misawa et al., 1993; Römer et al., 2000) and/or overexpression of plant-derived carotenoid biosynthesis genes (Paine et al., 2005; Ye et al., 2000). Efforts were directed towards increasing the flux through the pathway and/or enhancing the levels of specific carotenoid species. Although increased β-carotene synthesis can also been achieved by conventional nuclear transformation (Rosati et al., 2000), engineering the pathway via plastid transformation offers several potential advantages. First, introduction of additional copies of carotenoid biosynthesis genes by nuclear transformation has frequently resulted in epigenetic gene inactivation by co-suppression (reviewed by Fray and Grierson, 1993). Epigenetic effects are absent from plastids (Bock, 2001), making transgene expression highly stable and unsusceptible to epigenetic variation. Secondly, the chloroplast genetic system allows expression of multiple transgenes from operons (De Cosa et al., 2001; Lössl et al., 2005; Quesada-Vargas et al., 2005), which greatly simplifies engineering of complex metabolic pathways by transgene stacking. Finally, plastids and plastid transgenes are maternally inherited in tomato (and most other crops), providing increased biosafety by reducing the risk of outcrossing and unwanted transgene transfer via pollen. Although it has been shown recently that chloroplast genes can occasionally be transferred to the nucleus, the frequency of these events is low and, most importantly, escaped chloroplast transgenes, due to their prokaryotic-type expression signals, are not expressed in the nucleus (Bock, 2005; Huang et al., 2003; Stegemann et al., 2003). Natural rates of outcrossing in tomato have been determined in several studies, and results have varied from very low levels to 43%, with values below 4% prevailing (reviewed by Scott, 1992). This great variation is attributed to the strong dependency of cross-pollination rates in tomato on weather conditions (which, for example, can cause style elongation/stigma exertion), local insect pollinations, and the genotype of the neighbouring varieties (Scott, 1992).

Thus far, plastid transformation has been routine only in tobacco (Maliga, 2004). The improved protocol for tomato plastid transformation reported here allowed the generation of a total of 17 transplastomic lines with four different transformation constructs. This high transformation efficiency will facilitate the routine use of tomatoes as a model system for metabolic engineering in plastids of an important food crop with an edible fruit. The enhancement of pro-vitamin A synthesis in transplastomic tomatoes represents a promising example for successfully engineering a nutritionally important biochemical pathway in non-green plastids by transformation of the chloroplast genome, and is an encouraging step towards the application of plastid transformation technologies in food crops.

Experimental procedures

Plant material

Sterile tobacco (N. tabacum cv. Petit Havana) plants were grown on agar-solidified MS medium containing 30 g l−1 sucrose (Murashige and Skoog, 1962). Sterile tomato (S. lycopersicum cv. IPA-6) plants were obtained from surface-sterilized seeds germinated and grown on agar-solidified MS medium with 20 g l−1 sucrose. Homoplasmic transplastomic lines were rooted and propagated on the same media. Rooted homoplasmic plants were transferred to soil and grown to maturity under glasshouse conditions.

Cloning procedures

Plastid transformation vector pRB96 is a derivative of the previously described vector pRB94 (Ruf et al., 2001). An expression cassette for passenger genes was constructed by inserting an nptII gene driven by the atpI promoter and terminated by the rps16 3′ untranslated region (3′ UTR) into the polylinker of pRB94 (Figure 1). The plastid atpI promoter was cloned by PCR amplification from tobacco DNA using the primer pair PatpI5′ (5′-TTTTGAATTCGAGCTCTAGCTATATAAGAAATCCT-3′) and PatpI3′ (5′-TTTTCCATGGTGCCTTGCCCTCTGAAAAAA-3′). With the primer sequences, a SacI and an NcoI restriction site were introduced into the PCR product (underlined sequences). The rps16 3′ UTR was cloned by PCR amplification using the primer pair Prps16–5′ (5′-TTTTTCTAGAGAAATTCAATTAAGGAAAT-3′) and Prps16–3′ (5′-AGAACACGGAATTCAATGG-3′). With the Prps16–5′ primer sequences, an XbaI restriction site was introduced (underlined sequence) into the PCR product, whereas the Prps16–3′ sequence contains an endogenous EcoRI site that was used for subsequent cloning steps (underlined). The final expression cassette, consisting of the atpI promoter, the nptII coding region and the rps16 3′ UTR, was control-sequenced then transferred as a SacI/EcoRI fragment into vector pRB94 yielding plasmid pRB96 (Figure 1). The coding regions of all genes of interest were integrated into pRB96 as NcoI/XbaI fragments. The primers for PCR amplification of transgenes were as follows (NcoI and XbaI restriction sites are underlined).

Lycopene β-cyclase gene from P. blakesleeanus:

  • PLC5′: 5′-AAAACCATGGGACTGACTTATATGGAAGTAC-3′

  • PLC3′: 5′-AAAATCTAGATTAAGCGTGGGCACGGTCGG-3′

Phytoene synthase gene from P. blakesleeanus:

  • PPS5′: 5′-ACGCCATGGTCCATATCTACATCACCC-3′

  • PPS3′: 5′-TTTTTCTAGATTAAATGACAGTAAAGGCCT-3′

Lycopene β-cyclase/phytoene synthase fusion gene from P. blakesleeanus:

  • PLC5′/PPS3′

Lycopene β-cyclase gene from E. herbicola:

  • EcrtY5′: 5′-TTTTCCATGGGGGATCTGATTTTAGTCGGC-3′

  • EcrtY3′: 5′-TTTTTCTAGACTTTATCTCGTCTGTCAGGA-3′

The Phycomyces genes were cloned from amplified cDNA. A cloned lycopene β-cyclase gene (crtY) from E. herbicola (kindly provided by Drs Peter Beyer and Salim Al-Babili, University of Freiburg, Germany) was engineered to introduce the 5′NcoI and 3′XbaI sites using the above primers.

Transformation of chloroplasts

Young leaves from sterile tobacco or tomato plants were bombarded with plasmid DNA-coated 0.6-μm gold particles using a biolistic gun (PDS1000He; Bio-Rad, Munich, Germany) with an improved Hepta Adaptor (Mologen, Berlin, Germany; now marketed by Bio-Rad) that allows a more even bombardment of the target tissue than the standard gun. Primary spectinomycin-resistant tobacco lines were selected on RMOP regeneration medium containing 500 mg l−1 spectinomycin (Svab and Maliga, 1993; Svab et al., 1990). Spectinomycin-resistant tomato lines were selected on a modified MS medium (RMOL) containing 1 mg l−1 thiamine hydrochloride, 0.2 mg l−1 IAA, 3 mg l−1 6-benzylaminopurine (BAP) and 500 mg l−1 spectinomycin. Plastid transformants were identified by double-resistance tests on medium containing both spectinomycin and streptomycin (500 mg l−1 each; Bock, 2001). For each transgene, several independent transplastomic lines were subjected to three to four additional rounds of regeneration on RMOP/spectinomycin to obtain homoplasmic tissue. In an alternative protocol, regenerated shoots were transferred to phytohormone-free spectinomycin-containing (500 mg l−1) MS medium and propagated by stem cuttings (three successive rounds).

Isolation of nucleic acids and DNA gel-blot analyses

Total plant DNAs were isolated from fresh leaf tissue by a cetyltrimethylammoniumbromide (CTAB)-based method (Doyle and Doyle, 1990). Total cellular RNA was extracted using the peqGOLD TriFast reagent (Peqlab GmbH, Erlangen, Germany). RNA samples for cDNA synthesis were purified by treatment with RNase-free DNase I (Roche, Mannheim, Germany). For Southern blot analysis, DNA samples (5 μg total DNA) were digested with restriction enzymes, separated by gel electrophoresis on 0.8% agarose gels, and transferred onto Hybond nylon membranes (Amersham, Buckinghamshire, UK) by capillary blotting using standard protocols. Total cellular RNA samples were electrophoresed in formaldehyde-containing 1% agarose gels and blotted onto Hybond nylon membranes. Hybridizations were performed at 65°C in Rapid-Hyb buffer (Amersham) following the manufacturer's protocol. Hybridization probes were purified by agarose gel electrophoresis following extraction of the DNA fragments of interest from excised gel slices using the GFX PCR (DNA and Gel Band Purification) kit (Amersham). A 1257-bp StyI/PstI restriction fragment derived from the trnG/psbC region of the N. tabacum chloroplast DNA (Figure 1a) was used as an RFLP probe to verify chloroplast transformation. Alternatively, a 550-bp PCR product generated by amplification of a portion of the psaB coding region with primers P7247 (5′-CCCAGAAAGAGGCTGGCCC-3′) and P7244 (5′-CCCAAGGGGCGGGAACTGC-3′) was used. Transgene-specific probes for Northern blot analyses were generated by PCR amplification with the same primer pairs as used for cloning (Phycomyces constructs), and by excising the Erwinia lycopene β-cyclase coding region with NcoI and XbaI (Figure 1a).

cDNA synthesis and PCR

Purified DNA-free RNA from P. blakesleeanus (5 μg) was used in a 50-μl cDNA synthesis reaction. Reverse transcription of RNA samples was primed with oligo(dT) primer, and elongation reactions were performed with SuperScript III RNase H-free reverse transcriptase (Invitrogen, Paisley, UK) according to the manufacturer's instructions. Total cellular DNA or first-strand cDNAs were amplified in an Eppendorf thermal cycler using GoTaq Flexi DNA Polymerase (Promega, Mannheim, Germany) and gene-specific primer pairs. The standard PCR program was 30–40 cycles of 1 min at 94°C, 40 sec at 58°C, and 1–2.5 min at 72°C with a 10-min extension of the first cycle at 94°C and a 5-min final extension at 72°C.

Herbicide-tolerance assays

The herbicide 2-(4-chlorophenylthio)-triethylamine (CPTA) is a specific inhibitor of lycopene β-cyclases. CPTA was chemically synthesized as described previously (Schuetz and Baldwin, 1958). Correctness of the synthesis and product quality were verified by melting-point determination and nuclear magnetic resonance (NMR) analysis. Tolerance of transplastomic tobacco and tomato plants to CPTA was assessed by watering glasshouse-grown plants in pots with CPTA solution (50 or 100 μm) followed by phenotypic comparison with a water-treated control after 5 and 10 days, respectively. CPTA treatments of tomato fruits were conducted by harvesting fruits with pedicels before the breaker stage, followed by incubation of the pedicel (with the fruit attached) in CPTA solution (100 μm to 2.5 mm) until the ripening process was virtually complete.

HPLC analyses of pigments

Carotenoids and chlorophylls were isolated from dried leaf tissue by extraction with 80% acetone followed by a second extraction with 100% acetone and combination of the two extracts. Fruit carotenoids were isolated from dried tomato fruit tissue (harvested 20 days after breaker stage and analysed at the onset of fruit softening) using two-phase separation (phase 1: 50% hexane, 20% chloroform, 20% methanol, 10% acetone; phase 2: H2O). The organic phase was used for HPLC analyses. Separation, identification and quantification of carotenoids was performed by HPLC using an Agilent 1100 Series HPLC system with a diode array detection unit (Agilent, Waldbronn, Germany). All pigment species were quantified against known amounts of standards. For all separations, a YMC ODS-A 250 × 4.6 mm column + precolumn was used. Separation was performed as described by Thayer and Björkman (1990) with the following modifications: solvent A contained acetonitrile, methanol and 0.1 m Tris HCl pH 8.0 (72:8:3). Pigments were eluted using 100% solvent A for the first 5 min, followed by a shift to 100% solvent B for 20 min. The column was allowed to re-equilibrate in solvent A for 10 min prior to the next run.

Acknowledgements

We are grateful to Drs Peter Beyer and Salim Al-Babili (University of Freiburg, Germany) for helpful discussions and for an Erwinia lycopene β-cyclase clone. We thank Peter Dörmann (MPI-MP) for helpful suggestions for the measurements of carotenoid contents, Daniel Karcher, Irving J. Berger and Tercilio Calsa Jr for plasmid constructs, Marita Hermann and Stefanie Seeger for help with plant transformation, and the MPI-MP Green Team for plant care and cultivation. Seeds of various tomato varieties were kindly provided by Drs Dieter Scharf (Universit of Frankfurt, Germany), Helaine Carrer (University of Sao Paulo, Brazil) and John Gray (Cambridge University, UK). We are especially grateful to Dr Gerhard Erker and Peter Eggert (Organisch-Chemisches Institut, University of Münster, Germany) for help with the synthesis of CPTA and its NMR analysis, and to Dr Matthias Schroff (Mologen, Berlin, Germany) for technical improvement of our biolistic particle-delivery device. This research was supported by a grant from the Deutsche Forschungsgemeinschaft (BO 1482/11-1) to R.B. and by the Max Planck Society.

Ancillary