• Open Access

Generation and characterization of soybean and marker-free tobacco plastid transformants over-expressing a bacterial 4-hydroxyphenylpyruvate dioxygenase which provides strong herbicide tolerance

Authors


* Correspondence (fax +33 (0)4 72 85 22 97; e-mail ghislaine.tissot@bayercropscience.com)

Summary

Plant 4-hydroxyphenylpyruvate dioxygenase (HPPD) is part of the biosynthetic pathway leading to plastoquinone and vitamin E. This enzyme is also the molecular target of various new bleaching herbicides for which genetically engineered tolerant crops are being developed. We have expressed a sensitive bacterial hppd gene from Pseudomonas fluorescens in plastid transformants of tobacco and soybean and characterized in detail the recombinant lines. HPPD accumulates to approximately 5% of total soluble protein in transgenic chloroplasts of both species. As a result, the soybean and tobacco plastid transformants acquire a strong herbicide tolerance, performing better than nuclear transformants. In contrast, the over-expression of HPPD has no significant impact on the vitamin E content of leaves or seeds, quantitatively or qualitatively. A new strategy is presented and exemplified in tobacco which allows the rapid generation of antibiotic marker-free plastid transformants containing the herbicide tolerance gene only. This work reports, for the first time, the plastome engineering for herbicide tolerance in a major agronomic crop, and a technology leading to marker-free lines for this trait.

Introduction

Genetically modified herbicide-tolerant lines of soybean, canola, cotton and corn have been widely adopted by farmers worldwide, simplifying crop and weed management. All present commercial transgenic lines harbour the herbicide-tolerant gene in the nuclear genome. In most cases, the targeted pathways or enzymes occur in plastids (Freyssinet, 2003; Dhingra and Daniell, 2004). The generation of herbicide-tolerant plants via genetically engineered plastomes is also feasible. This has been demonstrated in tobacco (Nicotiana tabacum) only for tolerance to glyphosate (Daniell et al., 1998; Ye et al., 2001, 2003; Chin et al., 2003), phosphinothricin (Iamtham and Day, 2000; Lutz et al., 2001; Kang et al., 2003; Ye et al., 2003) and sulcotrione (Falk et al., 2005).

Herbicides such as sulcotrione and isoxaflutole (IFT) are inhibitors of 4-hydroxyphenylpyruvate dioxygenase (HPPD, EC 1.13.11.27). This enzyme is involved in the catabolism of aromatic amino acids (phenylalanine and tyrosine), catalysing the formation of homogentisate. In plants, this HPPD reaction product is a key precursor for the biosynthesis of photosynthetic pigments (such as plastoquinones) and vitamin E (tocopherols and tocotrienol derivatives), which are antioxidant compounds and elements of the photosynthetic electron transfer chain, respectively (Matringe et al., 2005). The enzymes involved in essential amino acid biosynthesis are encoded by nuclear genes, synthesized in the cytosol, and imported into plastids. Carrot and Arabidopsis HPPD have been localized in the cytosol, but all subsequent enzymatic steps in this biosynthetic pathway occur in plastids (Garcia et al., 1997, 1999).

IFT is a very potent herbicide for the selective pre- and early post-emergence control of a wide range of broadleaf and grass weeds in maize. IFT induces in susceptible species a characteristic bleaching of emerging treated foliar tissues. This effect is accompanied by a depletion of the available plastoquinone and vitamin E pools. The subsequent decrease in carotenoid levels causes foliage bleaching because the photosynthetic apparatus is no longer stabilized by these pigments. At high light intensity, excess energy is not quenched and chlorophyll molecules are destroyed. The inhibitor of HPPD is a derivative of IFT, diketonitrile (DKN), which is formed rapidly in plants following root and shoot uptake. In addition to agrochemical interest, other possible benefits expected from increasing the flux in this metabolic pathway are higher levels of vitamin E. As a consequence, the transgenic crops should present a better resistance to oxidative stress.

Plants tolerant to HPPD inhibitors have been engineered by over-producing from the nuclear genome a sensitive enzyme from Pseudomonas fluorescens (Sailland et al., 1996), Arabidopsis thaliana (Garcia et al., 1999; Rippert et al., 2004) and barley (Falk et al., 2003), or via plastid transformation in tobacco (Falk et al., 2005).

Plastid genetic engineering differs from nuclear transformation in many ways, as recently reviewed by Maliga (2004) and Daniell et al. (2004). The transgenes are integrated at a defined locus by homologous recombination, which simplifies drastically the screening of selected events because they are theoretically genetically identical. Because plastids are, in general, maternally inherited, the engineered traits do not segregate in the following generations via the male reproductive organ, although the probability of transmission by the pollen is not null. Plastids present several prokaryotic features reminiscent of their eubacterial origin (Margulis, 1975). The plastid genome, termed plastome, is a highly polyploid double-stranded circular DNA molecule that contains many clusters of genes organized in operons (Shinozaki et al., 1986; Bendich, 1987). The transcription and translation machineries are prokaryotic-like. Genes are transcribed by two distinct DNA-dependent RNA polymerases, encoded either by the nuclear genome (NEP) or by the plastome (PEP) (Hajdukiewicz et al., 1997; reviewed by Shiina et al., 2005). As a consequence, translation in plastids can be inhibited by antibiotics. Since the first report of tobacco plastid transformation (Svab et al., 1990), the aadA gene conferring spectinomycin and streptomycin resistance has been the most commonly used selectable marker for the production of plastid transformants. The recombinant protein expression level is often very high, at least in green tissues, owing to the very active chloroplast transcription and translation machineries, and the presence of hundreds to thousands of copies of the transgene per cell. These features have been largely exemplified in tobacco, and only very rarely in other species, notably crops, for which the selection of plastid transformants remains a technical challenge. Nevertheless, this situation is changing rapidly, with recent reports describing the plastid transformation of major agronomic crops, such as canola, although no stable lines were obtained (Hou et al., 2003), cotton (Kumar et al., 2004a) and soybean (Dufourmantel et al., 2004). It is therefore now possible to advance this technology beyond the proof of concept experiments performed in tobacco.

The removal of the antibiotic resistance gene after the generation of transgenic plants is desirable to eliminate the risk of antibiotic resistance gene flow, a public concern. Methods exist to remove the antibiotic resistance marker from transplastomic plants, including marker excision using the Cre/lox site-specific recombination system (Corneille et al., 2001; Hajdukiewicz et al., 2001; Lutz et al., 2006) and the generation of deletion derivatives after homologous recombination-mediated excision (Iamtham and Day, 2000; Klaus et al., 2004).

In this study, we describe the generation and detailed molecular characterization of plastid transformants of both tobacco and soybean over-expressing a bacterial hppd gene from P. fluorescens. The plastid transgenic lines of tobacco exhibit a strong increase in IFT tolerance vs. nuclear transformed plants. This is also the first report on herbicide tolerance engineered in plastid transformants outside tobacco. Moreover, we have also developed an original strategy, exemplified in tobacco, to generate antibiotic marker-free transformants using this hppd herbicide tolerance gene.

Results

Construction of tobacco and soybean plastid transformation vectors

The pCLT111 vector was designed to allow the targeted integration of the transgenes into the large single-copy region of the tobacco plastid genome, between the rbcL and accD genes (Figure 1a). The expression cassettes of the transgenes were flanked by two tobacco plastid DNA sequences: the left homologous recombination region (LHRR), containing the 3′ end of the rbcL gene, and the right homologous recombination region (RHRR), containing the 5′ part of the accD gene. The gene of interest was hppd from P. fluorescens. Its expression was controlled by tobacco plastid regulatory elements: the psbA promoter/5′ untranslated region (5′UTR) and the 3′ end of the rbcL gene (3′rbcL). The selection cassette was composed of the coding sequence of the aadA gene, which confers spectinomycin and streptomycin resistance, under the control of a chimeric tobacco promoter Prrn(p) containing only the binding site for the plastid-encoded polymerase fused to the ribosome-binding site (RBS) region from the tobacco rbcL gene. The aadA transcript was further stabilized by the 3′ end of the tobacco psbA gene, as described by Svab and Maliga (1993). The two transgene expression cassettes were in the same orientation as the resident adjacent rbcL and accD genes.

Figure 1.

(a) Integration of 4-hydroxyphenylpyruvate dioxygenase (hppd) and aadA genes into the wild-type (WT) tobacco plastid genome (A) after transformation with vector pCLT111, giving a transformed plastome (B). Following the recombination between the two 210-bp 3′rbcL, a deleted transformed plastome was obtained (C). Probes used for Southern analysis: aadA, HPPD, rbcL. (b) Integration of hppd and aadA genes into the WT soybean plastid genome (A) after transformation with vector pCLT323, giving a transformed plastome (B). Probes used for Southern analysis: HPPD, IR. LHRR and RHRR, left and right homologous recombination regions; RBS, ribosome-binding site region.

The pCLT323 vector was developed to integrate the expression cassettes into the inverted repeated region of the soybean plastid genome, between the rps12/7 and trnV genes (Figure 1b). The Glycine max LHRR and RHRR flanking the two expression cassettes have been described previously by Dufourmantel et al. (2004). The aadA selection cassette contained the same elements as in the tobacco pCLT111 vector. The expression of the hppd gene of P. fluorescens was under the control of plastid tobacco regulatory elements: the complete 16S ribosomal operon promoter Prrn(np) containing the binding sites for both the plastid and the nuclear-encoded RNA polymerases (PEP and NEP), the RBS region of the bacteriophage T7 gene 10 (G10L) (Ye et al., 2001) and, at the end of the hppd coding region, the 3′ end of the rbcL gene. The two expression cassettes were in the same orientation and transcribed in the opposite direction relative to that for the trnV and 16SrDNA resident genes.

Generation and analysis of tobacco and soybean plastid transformants

Leaves of N. tabacum (cv. PBD6) from in vitro-grown plantlets were bombarded with gold particles coated with the transforming vector pCLT111, essentially as described by Svab et al. (1990). Putative transformants were selected on medium containing 500 mg/L spectinomycin. All the antibiotic-resistant events were first screened by polymerase chain reaction (PCR) to check the insertion of the transgenes at the expected location in the tobacco plastome (data not shown). A second cycle of regeneration (R2) for some transformants of generation T0 was performed on medium containing spectinomycin and 1 mg/L DKN in order to promote the establishment of a homoplasmic state. The first generation of seeds, referred to as the T1 generation, from event numbers 111-1(R1), 111-9(R1) and 111-17(R1 and R2), was sown on medium supplemented with 20 mg/L DKN. The transformed events were further characterized by Southern blot analysis (Figure 2a). Total DNA from leaves was extracted from the spectinomycin-resistant event numbers 111-1, 111-4, 111-9, 111-17 and 111-25 of generation T0 (R1 or R2), from plants of generation T1 of the event 111-17 (R1 and R2) and from wild-type (WT) tobacco. The samples were digested by the restriction enzymes NcoI and HindIII, separated by agarose gel electrophoresis, blotted on to a nylon membrane and hybridized with [32P]-labelled probes specific for the rbcL, aadA and hppd coding regions, as depicted in Figure 1a.

Figure 2.

(a) Southern blot analysis of untransformed (wild-type, WT) and transformed (T0 and T1 from events 1, 4, 9, 17 and 25) tobacco plants. Total DNA digested by NcoI/HindIII was probed with chloroplast DNA flanking sequence containing a part of the tobacco rbcL gene (rbcL probe), with the 4-hydroxyphenylpyruvate dioxygenase gene (hppd) coding sequence (HPPD probe) and with aadA coding sequence (aadA probe). L, DNA ladder; pCLT111, transforming DNA digested by HindIII; R2, second cycle of regeneration. (b) Southern blot analysis of untransformed (WT) and transformed soybean T1 plants. Total DNA extracted from the initially selected callus (C), from leaves of four plants of generation T1 (A, B, C, D) and from wild-type soybean (WT), digested by HindIII, was probed with left plastid DNA flanking sequence (IR probe) and with hppd coding sequence (HPPD probe). L, DNA ladder.

A DNA fragment of 5923 bp was detected in all the lanes for generations T0 and T1 using the aadA probe, confirming the integration and stability of the transgenes at the predicted locus in the plastome (Figures 1a and 2a). The hppd probe revealed an expected fragment of 1296 bp, except for event 4. In this case, no signal was detected in the three analysed samples of three plants regenerated from event 4. An abnormal pattern was also detected for event 4 with the rbcL probe. Interestingly, in contrast with event 4, this band at around 1.8 kbp was very faint in most lines. The presence of such a fragment could have resulted from excision of the hppd gene, mediated by homologous recombination between the two 210-bp direct 3′rbcL repeats present in the native rbcL gene and in the herbicide resistance cassette (Figures 1a and 2a). Otherwise, the rbcL probe allowed the detection of a fragment of 2085 bp for events 1, 9, 17 and 25, as expected when the aadA and hppd transgenes are both integrated between the rbcL and accD plastome genes. A band at 6429 bp, specific for the WT plastome, was present in all lanes, suggesting that all these recombinant events are heteroplasmic. Another explanation, considering that the intensity of the signal was surprisingly uniform for all plants tested, is that the signal corresponds to nuclear copies of this plastid region, which are known to exist (Ayliffe and Timmis, 1992; Kode et al., 2006). No impact was noted on this pattern by a second round of regeneration under selective pressure, nor was any difference seen in the T1 progenies (Figure 2a).

Embryogenic calli of G. max (cv. Jack) were transformed by bombardment of gold particles coated with the transforming vector pCLT323, as described by Dufourmantel et al. (2004). Fourteen bombardments were performed, yielding only one spectinomycin-resistant event. Specific integration of the transgenes was confirmed by PCR (data not shown). The recombinant event was regenerated via somatic embryogenesis and further characterized by Southern blot analyses (Figure 2b). Total DNA extracted from the initially selected callus, from the leaves of one T0 plant and four plants of generation T1, and from WT soybean was digested by the restriction enzyme HindIII. After transfer on to a nylon membrane, the DNA fragments were detected by specific hybridization with [32P]-labelled probes covering the coding sequence of hppd and the non-coding intergenic region (IR) of the soybean plastid genome present in LHRR, as depicted in Figure 1b. The hppd probe revealed a fragment of 7740 bp in all T1 transgenic samples, confirming the integration of the two expression cassettes at the expected locus in the soybean plastome and their transmission to the progeny (Figure 2b). The IR probe hybridized a band of 3975 bp in the T1 transgenic samples. In this case, no signal corresponding to a WT fragment of 8756 bp was detected as in the non-transformed control. This analysis confirms that the homoplasmic state is reached rapidly, even at the callus stage, and maintained in the T1 generation.

Analysis of hppd expression

The accumulation of HPPD protein was examined in different tissues of transgenic tobacco plants (generation T0; event 111-17R2E). Total soluble proteins (10 µg) of mature seeds, leaves, petals and roots, as well as total soluble proteins extracted from leaves of WT tobacco, were analysed by Western blot (Figure 3a). Using polyclonal antibodies raised in rabbit against P. fluorescens HPPD, the recombinant protein could be detected at around the expected molecular weight of 40 kDa in seeds, leaves and petals. The level of accumulation was very high in leaves, estimated at around 5% of proteins, a consequence of the strong plastid metabolic activity in photosynthetic tissues relative to that in seeds or petals. The expression seen in mature tobacco seeds showed that recombinant proteins encoded by the plastome could also accumulate significantly in these non-green organs. No signal was detected in roots, in accordance with the use of the light-regulated promoter and 5′UTR from the psbA gene.

Figure 3.

(a) Accumulation of 4-hydroxyphenylpyruvate dioxygenase (HPPD) protein in different tobacco tissues of chloroplast engineered plants. Western blot analysis was performed with 10 µg of total soluble proteins extracted from transformed seeds, leaves, petals, roots and wild-type (WT) leaves. One hundred nanograms of standard HPPD protein were loaded and used as reference. (b) Accumulation of HPPD protein in chloroplast engineered soybean. Western blot analysis was performed with 40 µg of total soluble proteins extracted from leaves of plastid T0 transformants and WT soybean. A dilution series of standard HPPD protein was used for quantification (10–100 ng). (c) Two-dimensional polyacrylamide gel electrophoresis (2D PAGE) profile of HPPD protein accumulation in chloroplast engineered soybean. One hundred and fifty micrograms of total soluble proteins from leaves of plastid transformant plant expressing only the aadA gene (pCLT312; Dufourmantel et al. 2004) (A) and leaves from T0 plastid transformant plants transformed with aadA and hppd genes (pCLT323) (B) were loaded at constant protein concentration and separated by a pH 3–10 gradient used for isoelectric focusing. (C) The result of the sequencing by liquid chromatography/tandem mass spectrometry (LC/MS-MS) after trypsin fractionation of the large spot of recombinant HPPD (red circle) confirms the nature of the protein. Over 358 theoretical amino acids in HPPD protein; 212 amino acids were sequenced (in blue).

The accumulation of HPPD protein was studied in leaves of transgenic soybean plants of generation T0 (Figure 3b). Total soluble proteins (40 µg) were extracted from leaves of plastid transformants and compared with WT soybean. High expression of the HPPD enzyme was found in all four transgenic soybean plants, at a level similar to the results obtained in tobacco. The band of higher molecular weight visible on the blot at around 53 kDa is the large subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco).

Total soluble proteins from T0 transgenic soybean leaves were also separated by two-dimensional gel electrophoresis (sodium dodecylsulphate-polyacrylamide gel electrophoresis, SDS-PAGE; Figure 3c). A large, single spot was visible on Coomassie blue staining at approximately the expected position for HPPD in the transgenic sample (molecular weight, 40 kDa; pI = 5), which was not present at all in the WT extract. To confirm its identity and analyse the recombinant protein, the corresponding spot was isolated, fragmented with trypsin and the fragments were sequenced by liquid chromatography/tandem mass spectrometry (LC/MS-MS). The sequences obtained (Figure 3c; blue) confirmed the identification of the spot as HPPD, and provided the coverage of around 59% of the entire protein sequence (211 residues). This analysis showed no sign of any mutation or editing event in the sequenced HPPD fragments.

Herbicide tolerance and vitamin E content of tobacco and soybean transformants

The herbicide tolerance of transgenic tobacco (generation T1 of transformation events 1, 4, 9 and 111-17R2E) and soybean (generation T1) was evaluated under glasshouse conditions, and compared with the tolerance level engineered in the nuclear genome of both species. Two-month-old tobacco plants received a high-dose post-emergence spray equivalent to 1 or 2 kg/ha of IFT (Balance™), and the phenotype was scored 3 weeks later. At such herbicide levels, control PBD6 WT plants presented the classical symptoms of leaf bleaching associated with HPPD inhibitors (Figure 4a, 2 and 3). The T1 generation of event 4, with conservation of the aadA selectable marker gene only and not the hppd gene, presented the same phenotype (not shown). In contrast, even at the higher dose, no phytotoxicity was scored in the T1 generation of the three other transgenic events, as shown for 111-17R2E (Figure 4a, 5 and 6). The IFT tolerance was compared with that of the most tolerant nuclear transgenic line generated in our laboratory (A. Sailland, unpublished results). In this case, the tobacco cultivar was ‘Petit Havana’, and the hppd gene was placed under the control of the strong constitutive viral cassava vein mosaic virus (CsVMV) promoter (Verdaguer et al., 1996). The HPPD protein, expressed in the cytosol, was targeted to the plastids. Some phytotoxicity was noted at 1 kg/ha (Figure 4a, 8), and major leaf damage and stunting were evident at 2 kg/ha (Figure 4a, 9). There was therefore a striking advantage of the tobacco plastid transformants in terms of post-emergence tolerance to IFT, which was in agreement with the high-level expression of HPPD.

Figure 4.

(a) Tobacco herbicide tolerance of plastid transformants. Comparison of wild-type tobacco, nuclear and plastid transformants scored 3 weeks after isoxaflutole (IFT) treatment (1 kg and 2 kg of IFT). (b) Soybean herbicide tolerance of plastid transformants. Comparison of wild-type soybean, nuclear and plastid transformants scored 3 weeks after IFT treatment (160 mg/L).

Two-week-old soybean plants of generation T1 were drenched with a solution of IFT (160 mg/L) and scored 3 weeks after treatment (Figure 4b). The WT plants were completely bleached or necrotic. The most IFT-tolerant nuclear transgenic line of soybean showed important symptoms of phytotoxicity. In this case, hppd expression was under the control of the sunflower rbcS promoter and the HPPD protein was also targeted to the plastids (B. Pelissier, unpublished results). The plastid transformants showed no phenotypic alteration, and performed, as in tobacco, much better than the nuclear lines (Figure 4b).

The vitamin E content of tobacco and soybean plastid transformants was analysed and compared with that of WT plants. The analysis was also performed for the most tolerant nuclear transgenic tobacco and soybean lines mentioned above. In all cases, no significant quantitative or qualitative differences were found between transgenic and WT material in either seed or leaf extracts (data not shown).

Generation of marker-free, herbicide-tolerant tobacco transformants

Nuclear HPPD transformants of tobacco can be isolated by direct selection for DKN tolerance (on media containing 1 mg/L DKN). Attempts to use the hppd herbicide tolerance gene to select directly plastid transformants in tobacco were not successful, most probably because DKN, the active metabolite of IFT used in the in vitro experiments, induces only the bleaching of tissues and does not stop cell proliferation under these heterotrophic conditions.

Because the presence of an antibiotic resistance gene in plants is not desirable, we have developed a strategy to produce marker-free, herbicide-tolerant plants which relies on the plastid recombination machinery. The plastid vector pCLT146 targets the rbcL-accD insertion site and carries the aadA selection cassette as in pCLT111, but with the terminator of the tobacco rps16 gene (Figure 5). This cassette was inserted between two fragments of the HPPD coding region, corresponding to the NH2-terminus (segment HP1: 579 nucleotides) and COOH-terminus (segment P1PD: 901 nucleotides) of the protein. These two sequences overlap over a segment P1 of 403 nucleotides. If the two copies of P1 recombine inside the plastid genome, this will lead to the elimination of the aadA cassette and the restoration of the full-length HPPD coding region, which is placed under the control of the tobacco psbA promoter and terminator. Then, a functional HPPD enzyme will be produced, conferring DKN tolerance to the recombined plastid transgenic plant.

Figure 5.

Selectable marker elimination. Integration of transgenes into the wild-type (WT) tobacco plastid genome (A) after transformation with vector pCLT146, giving a transformed plastome (B). After the recombination between the two P1 repeats, a marker-free plastome was obtained (C). HP1, NH2-terminus fragment of the 4-hydroxyphenylpyruvate dioxygenase gene (hppd) coding region; LHRR and RHRR, left and right homologous recombination regions; P1, repeat segment overlapping the NH2- and COOH-terminus fragments; P1PD, COOH-terminus fragment of the hppd coding region. Probes used for Southern analysis: aadA, P1, rbcL.

Tobacco plastid transformants were generated classically by selection on spectinomycin and checked by PCR for the presence of the transgenes inserted inside the plastome. We focused the analysis on two representative events (T0; numbers 146-1 and 146-2) and their progenies. The T1 progenies of events 146-1 and 146-2 (not shown) were sown in vitro in the presence of DKN (1 or 10 mg/L) or spectinomycin (500 mg/L), and compared with the T1 progeny of line 111-17R2E obtained with pCLT111 (Figure 6). The tested progenies were all entirely resistant to spectinomycin, as expected for plastid transformants which have, in tobacco, as in most higher plants, a plastid maternal inheritance. The lines generated with pCLT146, as exemplified by the progeny of event 146-1, showed a significant level of herbicide tolerance. Plantlets were tolerant to 1 p.p.m. of DKN, a concentration for which the WT was completely bleached (not shown). At 10 mg/L of DKN, there was clearly more phytotoxicity than for line 111-17R2E, and a very high heterogeneity in pigmentation and tolerance, often in the same plantlets or within the same leaves (Figure 6). Plants showing the best tolerance to 10 mg/L of DKN were selected from T1 progenies of events 146-1 and 146-2 (named 146-1a, 146-1b, 146-1c, 146-2a, 146-2b, 146-2c) and grown in the glasshouse (generation T1DR1).

Figure 6.

Comparison of the in vitro tobacco herbicide tolerance of plastid transformants. Seeds from 146-1 (A) and 111-17R2E (B) lines germinated on 500 mg/L spectinomycin and on 1 mg/L of diketonitrile (DKN). Enlarged windows of the seedlings from 146-1 (C) and 111-17R2E (D) germinated on 10 mg/L of DKN.

In parallel, leaves of events 146-1 and 146-2 (T0) were dissected and used in vitro for a second round of plant regeneration (R2) in the absence of any selection agent, producing plantlets of generation T0R2. After 1 month, some plantlets were rooted on media containing 1 mg/L of DKN. Plants showing the best herbicide tolerance (146-1A and 146-2A) were transferred to the glasshouse. Seeds from these lines were sown in vitro on media containing 20 mg/L of DKN. The T1DR2 plants obtained (146-1Aa–d, 146-2Ab–d) were further analysed at the molecular level.

Southern blot analysis was performed on generation T0 from events 146-1 and 146-2, and on their T1 progenies (T1DR1 and T1DR2). Total DNA of untransformed (WT) and transformed plants was extracted from leaves and digested by SacI/XhoI. After transfer on to a nylon membrane, the DNA fragments were hybridized with [32P]-labelled probes specific for the aadA, rbcL and P1 coding sequences (Figure 7). The result on generation T0 shows that the entire construct was integrated at the expected locus in the tobacco plastome for both events (Figure 7). A band of weak intensity was detected at 5.7 kb with the P1 and rbcL probes, which corresponds to the size expected after recombination between the P1 repeats. For the six plants of generation T1DR1 (three from each event), which were selected for their good in vitro herbicide tolerance, the pattern was essentially the same, except for plant 146-1c which contained more recombined molecules at 5.7 kb. Finally, in generation T1DR2, plants were detected which had apparently completely eliminated the aadA marker cassette (no signal) and reconstituted the hppd gene, such as lines 146-2Ab and 146-2Ad. Other lines, such as 146-1Ad and 146-2Ac, had not completely eliminated the selectable marker.

Figure 7.

Southern blot analysis of untransformed (WT) tobacco, generation T0 from events 146-1 and 146-2 and their T1 progenies (T1DR1 and T1DR2). Total DNA digested by SacI/XhoI was probed with aadA coding sequence (aadA probe), with chloroplast DNA flanking sequence containing a part of the tobacco rbcL gene (rbcL probe) and with the overlapping P1 sequence (P1 probe). R2, second cycle of regeneration.

The T2 progeny of line 146-2Ab (T1DR2) were germinated in vitro in the presence of herbicide or spectinomycin (Figure 8a), and compared with generation T1 of event 146-2 and stable line 111-17R2E. It was found that the herbicide tolerance level of the T2 progeny to 10 mg/L DKN was much higher than that in generation T1, and similar (or even better) than that of line 111-17R2E. Moreover, this T2 line was susceptible to spectinomycin. In accordance with the Southern blot results and the phenotype, the aadA gene could not be recovered by PCR from this T2 generation (data not shown). The reconstitution of HPPD in the marker-free line 146-2Ab was examined on Coomassie Blue-stained membrane (Figure 8b). A major band was visible at the expected size for HPPD, and absent from WT. The confirmation of the identity of the protein was obtained by N-terminal sequencing after transfer on to a polyvinylidene difluoride (PVDF) membrane. The sequence of the first five residues corresponded exactly to HPPD, starting at the second position (ADLYE), in accordance with the rule that methionine is generally removed post-translationally when preceding an alanine (Hirel et al., 1989).

Figure 8.

(a) Phenotype of selected plastid transgenic lines. Seeds from generation T1 and T2 of 146-2 and from generation T1 of 111-17R2E germinated on 500 mg/L of spectinomycin and on 0, 1 and 10 mg/L of diketonitrile (DKN). (b) 4-Hydroxyphenylpyruvate dioxygenase (HPPD) accumulation in marker-free line 146-2Ab. Total soluble proteins (40 µg) extracted from leaves of the 146-2Ab line (lane 2) and wild-type tobacco (lane 3). Lane 1: standard HPPD (20 µg). The N-terminal sequence corresponding to the HPPD protein is shown on the Coomassie Blue-stained membrane.

Discussion

This is the first report to describe the genetic engineering of plastids for herbicide tolerance in a major crop. It also provides a comparison of the expression of a recombinant protein in plastid transformants of two different species: soybean and tobacco. As often observed with this technology, and for bacterial genes, high-level expression of HPPD was detected on Western blots in the green tissues of both species. The recombinant protein could also be visualized easily after total soluble protein separation on one or two dimensions on standard Coomassie Blue staining. We estimate that HPPD represents at least 5% of the soluble mature leaf proteins in transformed soybean and tobacco. It should be noted that these results in leaves are very similar in the two species, despite the use of two completely different sets of expression signals. The regulatory elements (the psbA promoter and its 5′UTR or the 16S rDNA promoter coupled to the RBS region of G10L) were chosen because of their previously described efficacy for driving the high expression of various recombinant proteins in chloroplasts (Ye et al., 2001; Fernandez-SanMillan et al., 2003). The soybean recombinant HPPD could be isolated from leaves after a two-dimensional separation and was characterized partially at the sequence level. The sequences perfectly matched the theoretical sequence over the 211 residues obtained. To our knowledge, this is only the second article, together with the publication concerning the expression of human growth hormone (Staub et al., 2000), providing sequence data for a recombinant protein expressed by plastid transformants. The important result is that there is no evidence for messenger editing of the introduced bacterial sequence, whereas around 30 sites are known to be edited for resident plastome genes in maize and tobacco (Maier et al., 1995; Hirose et al., 1999). Expression in other organs was examined in tobacco only, showing, not surprisingly, much smaller amounts of HPPD, especially in roots. The latter observation is consistent with the properties of the light-regulated tobacco psbA 5′UTR used to drive the expression of HPPD (Staub and Maliga, 1994). High expression in roots is nevertheless possible in plastid transformants, as recently demonstrated for carrot chromoplasts (Kumar et al., 2004b), although this might be a special context. The detection of HPPD in mature tobacco seeds, together with our previous report showing the presence of a recombinant Cry1Ab toxin in the beans of soybean plastid transformants (Dufourmantel et al., 2005), confirms the possibility of engineering plastids for traits which need to be expressed in this biologically and economically important organ. However, the precise tissue localization of the expression in the seeds (teguments, endosperm or embryo) needs to be determined.

One soybean transgenic event only was obtained from 14 bombardments. This transformation frequency is similar to that reported for the integration of the cry1Ab gene (one event for 11 bombardments) (Dufourmantel et al., 2005), and significantly lower than that obtained for the development of the technology in soybean (two events per bombardment) (Dufourmantel et al., 2004), using the same integration site and the same selection cassette. This decrease can probably be attributed to the intrinsic variability of this type of experiment involving in vitro tissue culture, as is the case in tobacco.

The post-emergence herbicide tolerance was evaluated for transgenic tobacco and soybean lines. Remarkably, no phytotoxicity was noted in tobacco plastid transformants treated at a dose of 2 kg/ha of IFT, which is far above the 100 g/ha agronomic treatment recommended for the eradication of weeds in the field. This tolerance level is also clearly superior to that observed in the most tolerant nuclear tobacco transgenic lines previously engineered in our laboratory (A. Sailland, unpublished results). This marked difference can be attributed directly to the increased expression level of HPPD, as nuclear transformants were also targeting the enzyme to the plastids. Herbicide treatment on soybean showed the same clear superiority of plastid engineered lines over nuclear transformants. It is noteworthy that the transgenic lines performed well despite the fact that plant HPPD is cytosolic in the species in which the localization has been examined (Lenne et al., 1995; Garcia et al., 1997).

The excellent tolerance to HPPD inhibitors of plastid transformants over-expressing the target enzyme has not been observed in tobacco lines engineered for glyphosate resistance (Ye et al., 2001). In this case, despite the high-level expression of the target enzyme (5-enolpyruvylshikimate-3-phosphate synthase, EPSPS) detected in tobacco leaves, the plastid transformants did not outperform the nuclear transformants targeting EPSPS to the plastids. Glyphosate, which inhibits the synthesis of aromatic amino acids, is toxic for all cell types, and recombinant EPSPS expression in some non-green tissues was probably too low in the plastid transformants. In contrast, inhibitors of HPPD, such as IFT, deplete the level of vitamin E and plastoquinones, and induce a characteristic bleaching of the tissues, but have no toxic effect on non-photosynthetic cells, or on in vitro heterotrophic cultures (Pallett et al., 1998). Plastid engineering for this trait is therefore particularly well suited, as only expression of the target enzyme in chloroplasts is needed. With regard to the resistance to phosphinothricin, an inhibitor of glutamine synthetase (Kishore and Shah, 1988), tobacco plastid transformants expressing a detoxifying enzyme were highly tolerant to the herbicide, but no comparison was made with nuclear transformants (Iamtham and Day, 2000; Lutz et al., 2001; Kang et al., 2003). Moreover, the high levels of bar and epsps genes in the plastids were not sufficient to allow direct selection of chloroplast transformants on medium containing herbicide (Lutz et al., 2001; Ye et al., 2003). Thus, the negative side of the lack of a toxic effect of IFT (or its active derivative DKN) on in vitro cultures is that we were unable to select directly plastid transformants with DKN, despite many attempts, and despite the fact that this is possible for nuclear engineering (Garcia et al., 1999). Nevertheless, in this report, we provide a method, exemplified in tobacco, to engineer plastid transformants containing only this herbicide tolerance gene.

The analysis of the vitamin E content of engineered tobacco and soybean transgenic plants revealed no significant alteration compared with WT plants. These results confirm our previous work on nuclear-transformed HPPD plants, showing that over-production of HPPD alone is not sufficient to increase the vitamin E content of transgenic plants (Garcia et al., 1999; Rippert et al., 2004). This increase requires the deregulation of the tyrosine flux at the level of TyrA (Rippert et al., 2004). Indeed, in plants, homogentisate, the aromatic precursor of vitamin E, derives from tyrosine via its keto-acid HPP, the substrate of HPPD. The present study reveals that, even in plants over-expressing HPPD at a very high level, the flux of homogentisate is not sufficient to deregulate the vitamin E content of these plants. Our results are in apparent contradiction with those recently reported by Falk et al. (2003, 2005), showing a slight increase in leaf or seed vitamin E content in tobacco plastid and nuclear transformants expressing barley HPPD.

We have successfully designed a strategy for generating antibiotic marker-free transformants based on the plastid natural homologous recombination machinery. Such methods are highly desirable, not only for the public acceptance of the technology, but also for the following reasons: (i) only few marker genes are available, which are essentially antibiotic resistance genes (aadA, nptII or aphA-6), betaine aldehyde dehydrogenase (BADH) potentially being the exception (Daniell et al., 2001); (ii) expression of the marker gene is an unnecessary metabolic burden for the recombinant plastids. A number of publications have reported various methods leading to plastid marker elimination: site-specific recombination systems, such as Cre/lox (Corneille et al., 2001; Hajdukiewicz et al., 2001; Lutz et al., 2006); a co-transformation approach of marker gene (aadA) and herbicide tolerance genes at two different plastome locations (Ye et al., 2003); and strategies relying on loop-out recombination between duplicated repeats in direct orientation (Fischer et al., 1996; Iamtham and Day, 2000; Klaus et al., 2004).

The recombination frequency between direct repeats cannot be controlled and is dependent on the number and length of the repeats (Iamtham and Day, 2000; Day et al., 2005). If this excision frequency is too high, the selection of homoplasmic marker-free plants may be a problem. If it is too low, the isolation of marker-free plants will not be possible, or will take too many generations. The length of the direct repeats is therefore critical in this type of strategy. In the tobacco lines generated with pCLT111, which have two directly repeated rbcL 3′ ends of 210 bp (in the resident gene and in the hppd cassette), we observed for most lines only a very faint band on the Southern blot (rbcL probe) which could correspond to the elimination of the hppd cassette. The faint signal did not increase between the T0 and T1 generation. Interestingly, one transformant was isolated (event 4) which had completely eliminated this cassette in the T0 generation. In this case, the recombination event occurred probably very early during the selection process. For the lines generated with pCLT146, the two copies of P1 flanking the aadA cassette create a 403-bp direct repeat, and were designed in such a way that loop-out recombination will reconstitute a functional hppd herbicide tolerance cassette. We anticipated that this would allow us to detect more easily marker-free plants, and eventually allow us to select for these events. Significant tolerance to DKN (1 mg/L) was observed in the T1 generation sown in vitro, confirming the Southern blot results. Higher concentrations of DKN (10 mg/L) made it possible to select apparently more tolerant plantlets, but this was correlated with loop-out recombined molecules in only one of the six plants analysed. An important step to recover aadA-free plants was to proceed to a second round of regeneration from leaves in vitro in the absence of antibiotic selection pressure. This allowed the recovery of herbicide-tolerant T0R2 plants and of aadA-free plants in the T1R2 progeny. Two of the seven T1DR2 plants analysed after this second round of regeneration were free of antibiotic marker, based on Southern blot analysis, and their progenies (T2 generation) were uniformly susceptible to spectinomycin. This was correlated with a high expression of HPPD, thus very precisely reconstituted by homologous recombination, and a strong herbicide tolerance. For glyphosate and phosphinothricin tolerance also, direct herbicide selection of plastid transformants is not possible, and is effective as a second step only, after amplification of recombinant molecules has occurred using the aadA selectable marker (Iamtham and Day, 2000; Ye et al., 2003). The selection step was carried out on T1 generation plantlets, but could be applied earlier, on the T0 generation, just after transgenic plant selection with spectinomycin. This marker elimination method is simple, efficient and particularly adapted to the engineering of traits that can be selected for, such as herbicide tolerance. The recent development of plastid transformation in major agronomic crops, such as soybean (Dufourmantel et al., 2004) and cotton (Kumar et al., 2004a), now makes it possible to consider seriously this genetic engineering technology as an alternative option to nuclear transformation.

Experimental procedures

Plastid transformation vectors

The tobacco plastid transformation vectors pCLT111 (DQ459069) and pCLT146 (DQ463359) allow the targeted integration of the transgenes between the rbcL and accD tobacco plastid genes. LHRR and RHRR correspond to tobacco plastome fragments 57 769–59 295 and 59 304–60 540, respectively (GenBank Z00044; Shinozaki et al. 1986).

In vector pCLT111, the P. fluorescens sequence encoding the 358-amino-acid HPPD enzyme (GenBank CQ830291) is placed under the control of the tobacco plastid psbA promoter and 5′UTR (Z00044; position 1598–1821 in reverse orientation) and the 3′ end of the rbcL gene (Z00044; position 59 036–59 246). The aadA selection cassette, described by Svab and Maliga (1993), is inserted downstream from the hppd cassette.

In vector pCLT146, the chimeric aadA gene differs at the 3′ end with the use of the rps16 terminator (Z00044; position 4933–5093). This selection cassette is inserted into the hppd coding region after nucleotide 579 and followed by the 177–1074-nucleotide 3′ fragment of the hppd coding region, disrupting the gene and its function. A 403-nucleotide fragment (‘P1’ in Figure 5) is thus duplicated in the same orientation. The 3′ end encoding HPPD in pCLT146 is followed by the 3′ end of the psbA gene (Z00044; position 142–535 in reverse orientation).

The vector pCLT323 (DQ459070) derives from pCLT327 (Dufourmantel et al., 2005). These vectors target the integration of the transgenes into a non-coding region of the G. max plastome, between the rps12/7 and trnV genes, in the inverted repeated region (GenBank NC007942; Saski et al., 2005). The coding sequence of the Bacillus thuringiensis toxin cry1Ab gene in pCLT327 was replaced by the coding sequence of the hppd gene of P. fluorescens.

Generation of tobacco and soybean plastid transformants

Nicotiana tabacum (cv. PBD6) plastid transformants were selected as described by Svab and Maliga (1993). A second regeneration cycle was performed for some lines, either with spectinomycin or with the herbicide DKN. G. max transformants (cv. Jack) were generated from embryogenic cultures as described by Dufourmantel et al. (2004).

DNA extraction, PCR and Southern blot analysis

Total plant DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Valencia, CA). PCRs were performed in a thermocycler using ReadyMix Taq PCR Reaction Mix (Sigma). The probes used for Southern blot analysis of tobacco, depicted in Figures 1 and 5, were amplified using the primer pairs: (i) 5′-GATCAACCTGATCCTCAACAACG-3′ and 5′-GAGCATTTCGTAATAAGTGTCTGG-3′ for hppd on lines generated with pCLT111; (ii) 5′-CAACAGCATCGCCTCCTACTTTGCG-3′ and 5′-TTCACGGAAGTTGAACAATTTCTCG-3′ covering the duplicated hppd region P1 in lines generated with pCLT146; (iii) 5′-GTAGAGCCGTTTATGAATGTCTTCG-3′ and 5′-AAGGATGTCCTAAAGTTCCTCCACC-3′ for rbcL; and (iv) 5′-GAAGCGGTGATCGCCGAAG-3′ and 5′-TTATTTGCCGACTACCTTGGTGATCTCGCC-3′ for aadA. For soybean, the probes used for Southern blot detection were amplified using the primer pairs: (i) 5′-CTAGTGGTACCGATCCAATCACGATCTTCTAATAAGAAC-3′ and 5′-GAACCTCCTTGCTTCTCTCATGTTACAATCCTCTTGCCGC-3′ for the IR of LHRR; and (ii) 5′-GAAGGAGATATACCCATGGC-3′ and 5′-CGTTGTTGAGGATCAGGTTG-3′ for hppd. The PCR fragments were purified using a PCR Purification Kit (Qiagen), and radiolabelled by random priming with a MEGAPRIME Kit (Qiagen).

Southern blot analyses were performed with 1 µg of total DNA extracted from soybean and 5 µg of total DNA extracted from tobacco, essentially according to Sambrook et al. (1989). DNA was digested with HindIII for soybean, NcoI and HindIII for tobacco lines generated with pCLT111, or SacI and XhoI for tobacco lines generated with pCLT146. For all experiments, the last most stringent wash was performed with a solution of 0.1 × sodium salt citrate (SSC) and 1% SDS at 65 °C.

Protein extraction and Western blot analyses

Tissues were frozen in liquid nitrogen and ground to a fine powder. Extraction buffer [tris(hydroxymethyl)aminomethane (Tris) 100 mm, dithiothreitol (DTT) 0.04%, glycerol 10%, protease inhibitors (Complete Mini Protease inhibitor cocktail tablets, Roche, Basel, Switzerland)] was added to the powder, and incubated on ice for 10 min. Total soluble proteins were recovered in the supernatant after a centrifugation step of 15 min at 20 000 g. Total soluble proteins from leaf, root, petal or seed cells were quantified using the Bio-Rad Protein Assay Reagent Kit (Bio-Rad, Hercules, CA), separated by 12% SDS-PAGE and transferred on to a nitrocellulose membrane (for soybean) or PVDF membrane (Millipore Corp., Bedford, MA) (for tobacco), using a liquid electroblotting apparatus (Mini-Protean 3 Cell, Bio-Rad). After 2 h of saturation at room temperature with TTBS buffer (Tris 100 mm, NaCl 0.9%, Tween 0.1%, pH 7.5) containing Western Blocking Reagent (Roche), membranes were washed three times with TTBS buffer and incubated for one night at 4 °C with TTBS buffer containing polyclonal antibodies raised in rabbit against P. fluorescens HPPD. After three washes with TTBS buffer, membranes were incubated for 2 h at room temperature with TTBS buffer containing secondary antibodies raised in goat against rabbit immunoglobulin Gs (IgGs), conjugated to alkaline phosphatase (Sigma). The membranes were revealed, after three washes with TTBS buffer and one wash with TBS buffer (Tris 100 mm, NaCl 0.9%), with the Immun-Star™ AP Substrate Pack (Bio-Rad) on Hyperfilm™ ECL (Amersham Pharmacia Biotech, Freiburg, Germany).

Vitamin E analyses

Freeze-dried leaves of 2-month-old tobacco and soybean plants and mature seeds were ground in liquid nitrogen; 150 mg was extracted three times with 1 mL of CH3OH under dim light and in the presence of argon in order to prevent the oxidation of vitamin E. The resulting supernatants were pooled and stored at −80 °C before analysis. For tocopherol and tocotrienol determinations, 100 µL of each sample was injected through a C18 high-performance liquid chromatography (HPLC) column (Spherisorb ODS2 250 × 4.6 mm, 5 µm; Interchim, Montluçon, France). Vitamin E was detected by fluorescence using excitation at 290 nm and recording the emission at 325 nm. Tocopherols and tocotrienols were quantified by comparison with standards (Merck, Haar, Germany). The HPLC system consisted of two 510 HPLC pumps and a 712 WISP autosampler (Waters, Manchester, UK), an HPLC UV detector and an SFM 25 fluorescence detector (Kontron Instruments, Eching, Germany). The flow rate was 1 mL/min with CH3OH–H2O (96 : 4, v/v) as solvent.

Two-dimensional gel electrophoresis and mass spectrometry

Total soluble proteins (150 µg) extracted from leaves of T0 soybean plastid transformants were first separated by electrophoresis according to charge, using gel strips forming an immobilized non-linear pH gradient from 3 to 10 (Immobiline DryStrip pH 3–10 NL, 18 cm; Amersham Biosciences). Strips were rehydrated for 14 h at 22 °C with thiourea–urea lysis buffer containing 2% (v/v) Triton X-100 and 20 mm DTT. Isoelectric focusing was performed at 22 °C in a Multiphor II system (Amersham Biosciences) for 1 h at 300 V and 7 h at 3500 V.

Proteins were separated a second time according to their molecular weight. Prior to the second dimension, the gel strips were equilibrated for 2 × 20 min in 2 × 100 mL equilibration solution containing 6 m urea, 30% (v/v) glycerol, 2.5% (w/v) SDS, 0.15 m Tris and 0.1 m HCl (Gorg et al., 1988; Harder et al., 1999; Job et al., 2005). DTT (50 mm) was added to the first equilibration solution and iodoacetamide (4%, w/v) to the second (Harder et al., 1999). Equilibrated gel strips were placed on top of vertical polyacrylamide gels [10% (v/v) acrylamide, 0.33% (w/v) piperazine diacrylamide, 0.18 m Trizma base, 0.166 m HCl, 0.07% (w/v) ammonium persulphate, 0.035% (v/v) Temed]. A denaturing solution [1% (w/v) low-melting agarose (Life Technologies/GibcoBRL, Cleveland, USA), 0.4% (w/v) SDS, 0.15 m BisTris and 0.1 m HCl] was loaded on to the gel strips. Electrophoresis was performed at 10 °C in a buffer (pH 8.3) containing 25 mm Trizma base, 200 mm taurine and 0.1% (w/v) SDS for 1 h at 35 V and 14 h at 110 V (Job et al., 2005). Proteins were revealed with Coomassie Blue GelCode stain reagent (Pierce, Rockford, IL). LC/MS-MS analysis of recombinant HPPD was performed by Dr Maya Beghazi (Unité Mixte de Recherche N°6175, Service de Spectrométrie de Masse pour la Protéomique, Institut National de la Recherche Agronomique, Nouzilly, France). The spots of interest were excised from two-dimensional SDS-PAGE gels with sterile tips and placed in 1.5-mL sterile tubes. Spots were then rinsed, reduced with 10 mm DTT, alkylated with 55 mm iodoacetamide and incubated overnight at 37 °C with 12.5 ng/µL trypsin (Sequencing grade, Roche) in 25 mm NH4HCO3, as described previously (Shevchenko et al., 1996). The tryptic fragments were extracted, dried, reconstituted with 2% acetonitrile (v/v) and 0.1% formic acid, and sonicated for 10 min. Analysis of tryptic peptides by MS-MS was performed on a nano-electrospray ionization quadrupole time-of-flight hybrid mass spectrometer (Q-TOF Ultima Global, Waters Micromass, Manchester, UK) coupled with a nano-HPLC (Cap-LC, Waters). The samples were loaded and desalted on a C18 precolumn (LC-Packings PepMap C18, 5 µm, 100 Å, 300 µm × 5 mm; LC-Packings, Amsterdam, Netherlands) at a flow rate of 20 µL/min isocratically with 0.1% formic acid. The peptides were separated on a C18 column (Atlantis dC18, 3 µm, 75 µm × 150 mm Nano Ease, Waters). After washing with solvent A (water–acetonitrile 98 : 2, v/v, 0.1% formic acid), a linear gradient from 5% to 60% of solvent B (water–acetonitrile 20 : 80, v/v, 0.1% formic acid) was developed over 80 min at a flow rate of 180 nL/min. The Q-TOF spectrometer was operated in the Data Dependent Analysis mode using a 1-s MS survey scan on three different precursor ions. The peptide masses and sequences obtained were either matched automatically to proteins in a non-redundant database (National Center for Biotechnology Information, NCBI) using the Mascot MS/MS Ions Search algorithm (http://www.matrixscience.com) or blasted manually against the current databases.

Acknowledgements

Part of this work was funded under the European Union Fifth Framework Programme Plastid Factory, grant number QLK-CT-1999-00692, and supported by RhoBio.

Ancillary