Protein trans-splicing in transgenic plant chloroplast: Reconstruction of herbicide resistance from split genes

Abstract

Inteins are intervening protein sequences that undergo self-excision from a precursor protein with concomitant joining of the flanking sequences. Here, we demonstrate intein trans-splicing in Nicotiana tabacum chloroplasts by using the naturally split Ssp DnaE intein. Trans-splicing occurred whether both intein fragments were encoded in the chloroplast or were separated into the chloroplast and nuclear genomes. A biolistic approach was used to integrate two fusion genes, one encoding aminoglycoside-3-adenyltransferase (aadA) and the first 123 aa of the Ssp DnaE intein (In) and the other encoding 36 C-terminal amino acid residues of the Ssp DnaE intein (Ic) and soluble modified green fluorescent protein (smGFP) into N. tabacum plastids. Expression of these gene fragments in the chloroplast resulted in ligated aadA-smGFP due to In-Ic-mediated trans-splicing. Furthermore, an N-terminal portion of the herbicide resistance gene 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) containing a chloroplast localization signal fused to In (EPSPSn-In) was integrated into the nuclear DNA of N. tabacum by using Agrobacterium tumefaciens-mediated transformation. The remaining EPSPS gene fragment (EPSPSc) fused to Ic (Ic-EPSPSc) was integrated into the chloroplast genome by homologous recombination. Western blot analysis of cell extracts from these plants showed a full-length EPSPS, demonstrating that the EPSPSn-In gene product migrated to the chloroplast and underwent trans-splicing. Furthermore, these transgenic plants displayed improved resistance to the herbicide N-(phosphonomethyl)glycine (glyphosate) when compared with wild-type N. tabacum.

Inteins are intervening protein sequences that undergo an autoexcision reaction, protein splicing, that results in ligation of the flanking protein regions, termed exteins (1). Inteins are present in the genome of eubacteria, eukaryota, and archaea, and to date, >100 have been identified (InBase: www.neb.com/neb/inteins.html; ref. 2). The blue-green algae Synechocystis sp. PCC6803 has four inteins that are encoded within the dnaE, dnaB, dnaX, and gyrB genes. The dnaE gene encodes a split copy of the replicative DNA polymerase III catalytic α subunit (DnaE), with the split portion separated by 745 kb of genomic DNA. The dnaE-n gene encodes the first 774 aa of DnaE and the dnaE-c gene encodes the remaining 423 aa of the DnaE protein. These two gene portions are transcribed from opposite strands of the genome. When translated, this split gene yields one polypeptide containing the N terminus of the DnaE (N-extein) fused to an intein fragment and a second fusion peptide containing the complementary intein fragment and the C terminus (C-extein) of the polymerase. The intein fragments of the fusion proteins assemble, then mediate a protein trans-splicing reaction, resulting in cleavage of both peptide bonds at the intein/polymerase junction and ligation of the flanking DnaE regions, giving rise to a mature, catalytically active DNA polymerase III catalytic α subunit (3, 4).

The intein portions of the Synechocystis sp. PCC6803 fusion dnaE gene products (Ssp DnaE intein) consist of two fragments of 123 (In) and 36 (Ic) aa. The fragments have homology to the N- and C-terminal splicing domains, respectively, of other intein family members. Unlike artificially split inteins (5–7), the Ssp DnaE intein fragments In and Ic are able to trans-splice in vitro without urea treatment (8). These inteins are also able to cyclize and trans-splice proteins in Escherichia coli (8, 9). In E. coli, in vivo protein trans-splicing has been demonstrated by using acetolactate synthase II (ALSII) as a model (10). The enzymes ALSI, II, and III, in both bacteria and plants, are targets for sulfonurea herbicides (11). ALSI and ALSIII, but not ALSII, are sensitive to feedback inhibition by valine. An artificial construction that split the ALSII gene was able to rescue an ALSII mutant in E. coli (10) by using protein trans-splicing in the presence of valine. Recently, we have demonstrated that a splicing-defective Ssp DnaE split intein could be used as an affinity domain for intein-mediated complementation of 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) in aroA mutant E. coli (12). Thus, we speculated that split genes could be integrated into different genomes of a cell to reduce/prevent gene transfer by way of pollen because organellar genomes such as chloroplasts are normally transmitted maternally in commercially important plant species.

In this study, we have demonstrated that protein trans-splicing with the Ssp DnaE intein can occur in plant chloroplasts. The genes encoding the trans-splicing protein fragments do not need to be located in the same organellar or nuclear genome. Two genes encoding fragments of a herbicide-resistant form of EPSPS were placed into separate genomes, the nuclear and chloroplast genome. Full-length EPSPS protein was generated in the chloroplast after translocation of the N-terminal EPSPS-intein fusion protein fragment to the plastid followed by trans-splicing. Gene flow from the above-mentioned transgenic plants to wild or weedy relatives would transmit only a portion of the full-length gene, correlating to reduced environmental impact. Also, the activity of protein-splicing elements in plant cells opens the possibility of protein engineering, such as intein-based protein cyclization or polymerization, in plant tissues.

Materials and Methods

Genes and Germplasm.

The soluble modified GFP (smgfp) gene was obtained from the Arabidopsis Biological Resource Center at Ohio State University (Columbus), and the glyphosate-resistant Salmonella typhimurium aroA gene with a P101S mutation was obtained from the American Type Culture Collection. The spectinomycin resistance-conferring aadA gene was obtained from I. Schildkraut (New England Biolabs). All E. coli strains and consumables were from New England Biolabs. Tobacco (Nicotiana tabacum, Xanthi, Smith) seeds were obtained from V. Sisson (North Carolina State University, Oxford).

Construction of Expression Vectors.

Polymerase chain reactions were used to amplify, clone, and construct fusion genes between aadA, In, smgfp, Ic, and epsps. The sequences of the primers along with underlined restriction enzyme sites are as follows: aadA 5′ GCCTTAATTAACCATGAGGGAAGCGGTGATCGCCG and 3′ TGCGGTCGACTTTGCCGACTACCTTGGTGATCTC; In 5′ AGGGAATTCGTCGACAAAT TTGCTGAATATTGCCTGTCT and 3′ GGCCTCGAGTTATTTAATTGTCCCAGCGTCAAGTAATG; smgfp 5′ CCCAAGCTTGGCGCCATGAGTAAAGGAGAAGAACTTTTCAC and 3′ GCGACCGGTTTATTTGTATAGTTCATCCATGCCATG; and Ic 5′ AGCTTTGTTTAAACCATGGTTAAAGTTATCGGTCGTAGATC and 3′ CAGCGTCGACGGCGCCGTGGGATTTGTTAAAGCAGTTAGCAGC.

Amplified aadA gene product was digested with PacI and SalI and the amplified In gene was digested with SalI and XhoI. After reconstruction of the aadA-In gene fusion, the product was ligated into the chloroplast gene-targeting vector p226 at PacI and XhoI sites or was cloned into E. coli expression vector pIH976 (P. Riggs, New England Biolabs). The E. coli expression vector was termed pIHaadA-In. Similar techniques were followed for the Ic-smgfp fusion construct. Amplified Ic gene was digested with PmeI and NarI. Amplified DNA of smgfp was digested with NarI and AgeI. After reconstruction, the fusion gene Ic-smgfp was ligated into the PmeI and AgeI sites of p226 or pAGR3 (W. Jack, New England Biolabs). Plasmid pAGR3 containing Ic-smgfp is referred to as pAGIc-GFP. The final chloroplast gene-targeting vector was p226ag containing both aadA-In and Ic-smgfp. Chloroplast transfer vector p226 was based on a pLITMUS 28 (New England Biolabs) backbone. This plasmid had tobacco PpsbA and TpsbA sequences in opposite orientations to facilitate expression of two different products simultaneously. A 265-nucleotide SbfI fragment from the lambda genome separates the two PpsbA promoters. A detailed description of this vector is available on request.

For nuclear transformations, the Agrobacterium tumefaciens binary vector pBI121 was used. The S. typhimurium aroA gene and a signal peptide sequence from petunia EPSPS were amplified to generate the EPSPSn-In and Ic-EPSPSc fusion constructs by using the following primers: petunia chloroplast localization signal peptide sequence 5′ TCCCCCGGGGCCATGGCACAAATTAACAACATGGCTC and 3′ GGCAGCTCTTCGAAGCACTATCTCAGAAGGCTTCTGTGC; and EPSPSn 5′ GGCAGCTCTTCCCTTCAACCCATCGCGCGGGTCGATGGCGCC and 3′ ACGGTCGACACCTGGAGAGTGATACTGTTGACC. Signal sequence from petunia EPSPS cDNA was amplified and ligated to the S. typhimurium EPSPSn fragment, which in turn was ligated to the In fragment to constitute a tripartite fusion gene. A tobacco codon-optimized EPSPSn was also made for this fusion construct. The three-part gene fusion was inserted into pBI121 to make pBIEPSPSn-In. We have also constructed and transformed the petunia EPSPSn and EPSPSc gene fragments into tobacco by using a similar approach.

For chloroplast transformations, EPSPSc fragment was excised from pEPS65 by using NarI and AgeI and ligated into the spectinomycin marker vector p226alg to obtain palgIcEPC. Plasmid p226alg has an Ic gene, thus making a fusion Ic-EPSPSc gene. The chloroplast gene transfer vector palgEPC contained all of the flanking sequences of p226 and was capable of homologous recombination.

Tobacco Chloroplast Transformation.

A biolistic PDS-1000 He particle delivery system (Bio-Rad) was used for chloroplast transformation. Gold particles (30 mg, 0.7 μM) were precipitated in a 1.5-ml Eppendorf tube. The pellet was suspended with 1 ml of 70% ethanol and 1 ml of sterile water. Water was removed after gold particles were precipitated by centrifugation. Particles were resuspended with 5 μl of DNA (1 μg/μl). Subsequently, 50 μl of freshly prepared 2.5 M CaCl₂ and 20 μl of 0.1 M spermidine (free base, tissue culture grade) was added to DNA/gold particle mixtures. The mixture was vortex mixed and centrifuged for 2–3 s, and the supernatant was removed. The pellet was washed with 70% ethanol and the particles were suspended in 48 μl of 100% ethanol. Healthy leaves from 4- to 5-week-old plants were placed on TSMCK [1× Murashige and Skoog (MS) basal salts/1× B5 vitamin mix/0.005 mg/liter kinetin/4 mg/ml p-chlorophenoxyacetic acid] media overnight at 28°C in darkness. A volume of 8 μl of DNA particle suspension was loaded into the biolistic apparatus, and each plate was bombarded at 1,300 psi (1 psi = 6.89 kPa) of helium. After bombardment, the leaves were placed in the light at 28°C for 2 days. After 2 days, bombarded leaves were cut and transferred to MST5 media [premixed MS media with sucrose and agar (Sigma), 0.1 mg/liter α-naphthaleneacetic acid, and 1 mg/liter 6-benzylaminopurine supplemented with 500 mg/liter spectinomycin] for first selection. After 2 weeks, those sections were transferred to the fresh MST5 medium. Within 2–3 weeks, tiny green spots appeared and multiple shoots emerged from those sectors. Shoots (≈10–15 mm long) were placed in MS medium for rooting in Magenta boxes.

Nuclear Transformation.

A. tumefaciens strain LBA4404 was electroporated with pBI121EPSPSn-In, and three colonies were grown on LB at 30°C and 150–250 rpm overnight. Cultures were diluted 1:1 with LB and allowed to grow to an A₅₅₀ of ≈1.0. For cocultivation, leaf sections from 4- to 5-week-old tobacco shoot cultures were used. After the leaf sections were treated with A. tumefaciens, they were placed on a cocultivation medium (1× MS salts/3% sucrose/2 mg/liter α-naphthaleneacetic acid/0.5 mg/liter 6-benzylaminopurine) and incubated at 28°C in darkness for 2–3 days. Leaf sections were selected on medium containing MS salt, 3% sucrose, 0.5 mg/liter 6-benzylaminopurine, 500 mg/liter carbenicillin, and 100 mg/liter kanamycin. For sequential transformation, pBI121EPSPSn-In was transformed first, and F₁ plant leaves were subjected to chloroplast transformation by using a biolistic approach.

Southern Blot Analysis.

Genomic DNA was isolated from plant leaves by using the DNeasy kit (Qiagen, Valencia, CA). The purified DNA was digested with EcoRI, and the digested DNA fragments were separated on an 0.8% TBE (89 mM Tris/89 mM boric acid/2 mM EDTA, pH 8.3) agarose gel. The DNA in the gel was blotted onto Hybond N+ (Amersham Biosciences) and probed with a random-primed gfp or right targeting region DNA probe. The blot was washed as described (13) and autoradiographed.

Western Blot Analysis.

Cellular proteins from plants were extracted by using EB (50 mM Tris⋅HCl, pH 8.0, containing 200 mM NaCl, 5 mM EDTA, and 0.1% Tween 20). Extracts were mixed with SDS loading dye containing DTT, boiled at 95°C for 5 min, and loaded on a 4–20% or 10–20% Tris-glycine gradient gel (NOVEX, Carlsbad, CA). The proteins were blotted onto an Immobilon-P/Nitrocellulose membrane and probed with anti-intein/EPSPS antibodies raised in rabbit, followed by chemiluminescent detection (Cell Signaling Technology, Beverly, MA).

Microscopy and GFP Detection.

Protoplasts were made from transplastomic plants by using cellulase Onozuka R10 and macerozyme R10 (Yakult Pharmaceutical, Tokyo) and viewed with a Zeiss digital camera on a Nikon fluorescence microscope under a coverslip with UV light.

Results

In Vivo Trans-Splicing of Aminoglycoside-3-Adenyltransferase and smGFP in E. coli.

Aminoglycoside-3-adenyltransferase encoded by aadA confers resistance to spectinomycin. To examine whether trans-splicing of two unrelated proteins, such as aminoglycoside-3-adenyltransferase and smGFP, is feasible, we made two constructs that could be coexpressed in E. coli. Construct pIHaadA-In contained essentially the whole aadA gene without its signal peptide sequence, fused to DNA encoding amino acids 1–123 of In at its 3′ end. This construct was driven by the tac promoter and had a pACYC184 origin of replication along with a tetracycline-resistant marker. The second construct, pAGIc-GFP, contained DNA sequences encoding 36 aa of Ic fused to the coding sequence for a soluble version of GFP. The presence of a pBR322 ori and ampicillin-resistant marker gene in the second plasmid (Fig. 1A) ensured compatible maintenance and expression of the two plasmids in E. coli. Transformants containing both plasmids were able to grow on media supplemented with spectinomycin, ampicillin, and tetracycline, suggesting functional activity of the aadA-In fusion gene product (data not shown). To examine whether trans-splicing of the two unrelated gene products aadA and smGFP occurred, we performed a Western blot assay using the crude cell extract from transformed E. coli containing pIHaadA-In, pAGIc-GFP, or both by using anti-GFP monoclonal antibodies. A trans-spliced protein between aadA and smGFP was observed at the size expected for the fusion (57 kDa) only in cell extracts from E. coli ER1992, in which aadA-In and Ic-smGFP were coexpressed (Fig. 1B), demonstrating trans-splicing of unrelated proteins in E. coli.

Figure 1.

Protein trans-splicing in vivo. (A) Schematic diagram depicting trans-splicing of aadA and smGFP. Expression constructs: pIHaadA-In contains aadA-In fusion gene, whereas pAGIc-GFP contains Ic fused with smGFP. Fusion protein aadA-In and Ic-smGFP are ≈44 and 31 kDa, respectively. After the trans-splicing event a 57-kDa aadA-smGFP fusion protein will be observed. Intein fragments are excised. (B) Trans-splicing in E. coli. Western blot showing trans-spliced aadA-smGFP of ≈57 kDa as shown by an arrow in cell extract when pIHaadA-In and pAGIc-GFP were coexpressed in vivo. Anti-GFP monoclonal antibody was used for the detection. In the lane containing pAGIc-GFP, transformed cell extract shows only Ic-smGFP at 31 kDa. A cross-reacting band of ≈42 kDa was also visible.

Trans-Splicing of Aminoglycoside-3-Adenyltransferase and smGFP in Tobacco Chloroplasts.

A chloroplast transformation approach was undertaken to examine whether protein trans-splicing would take place in an organellar environment. Transfer vector p226ag containing aadA-In and Ic-smGFP fusion DNA sequences was chosen. Both fusion genes were under the control of the psbA promoter (PpsbA) and terminator (TpsbA). In tobacco PpsbA is shown to be highly active in the chloroplast environment (14). The two fusion genes were transcribed in opposite directions to obviate readthrough, and the promoters were separated by 265 nucleotides. The cassette had flanking chloroplast DNA sequences from 16S rDNA/trnaV loci (1,680 bp) as the left targeting region and rps12/7 loci (1,310 bp) as the right targeting region (Fig. 2A Top). This construct is capable of homologous recombination with the chloroplast genome of tobacco (Fig. 2A Middle). After recombination, the transgenic plastid will acquire the DNA between 16S rDNA/trnaV and rps12/7 (Fig. 2A Bottom). Because the PpsbA promoter is active in E. coli, we first examined whether fusion genes encoded by p226ag will be trans-spliced in E. coli. Western blot analysis of the transformed E. coli cell extracts demonstrated that the majority of the Ic-smGFP indeed trans-spliced with aadA as evidenced by the expected band of 57 kDa that reacted with the anti-GFP monoclonal antibodies (Fig. 2B).

Figure 2.

Conceptual gene targeting, homologous recombination in chloroplast genome, and expression of chloroplast gene-targeting construct in E. coli. (A) P and T represent the promoter and terminator of psbA. Left and right targeting regions are 16S rDNA/trnV (1,680 bp) and rps12/7 loci of the tobacco chloroplast genome, respectively. Restriction sites are indicated on top, and the fragments generated by EcoRI are indicated. (B) Plant promoter PpsbA expressed aadA-In and Ic-smGFP in E. coli, and the trans-spliced protein band was ≈57 kDa, as indicated by an arrow. Molecular mass markers are on the left and construct names are on the top of the gel. p226ag1–4 had both aadA-In and Ic-smGFP, but p226g1 contained only Ic-smGFP. Control represents the empty vector. Anti-GFP monoclonal antibody was used for detection.

To test whether protein trans-splicing can occur in the higher plant chloroplast, tobacco leaves were used for biolistic-mediated gene transformation. Several healthy tobacco leaves were transformed with the p226ag construct. After three to four selections on MST5 media, several putative transplastomic plants were obtained. To verify the correct recombination, transplastomic plant DNA was isolated and Southern blot and PCR analyses were performed. Two probes were used for Southern blot analysis, (i) the right targeting region of the transfer plasmid, p226ag, containing the whole rps17/27 loci and (ii) gfp. DNA was digested by EcoRI, which lacks recognition sequences within the transgene as well as the targeting regions (Fig. 2A) but cuts at least one site in close proximity to the 5′ region of 16S rDNA/trnaV and the 3′ region of the rps12/7. The DNA from one control and two transplastomic plants (Np226c4 and Np226c6) was subjected to Southern blot analysis by using both probes (Fig. 3 A and B). The right targeting region probe detected a band of ≈3.2 kbp in the control plant, whereas the transplastomic lines had 6.2-kbp bands, demonstrating homologous integration of the expected region of the transfer plasmid. This observation was confirmed further by using a gfp probe on the same set of DNA. Because tobacco does not contain gfp, only the transplastomic lines showed hybridization to the probe, yielding an ≈6.2-kbp band of identical size to the previous probe.

Figure 3.

Homologous recombination, phenotype, and protein expression in transplastomic tobacco plants. Southern blot analysis of the genomic DNA of transplastomic plants, Np226c4 and Np226c6, along with wild-type control tobacco plants probed with either the right targeting region (A) or gfp (B). (C) Representative transplastomic plant under white light. (D) Same plant as that in C under UV light. (E) Control wild-type plant protoplast under UV light. (F) Transplastomic plant (Np226c4) protoplast under UV light. (G) Trans-spliced aadA-smGFP protein, 57 kDa calculated molecular mass, as indicated by an arrow, from transplastomic plants. Plants are indicated on the top. Antibody used for detection is indicated at the bottom. (H) Trans-splicing in vivo. Plant extract with (+) or without (−) inhibitor Zn²⁺. Precursor and spliced products are indicated. Antibody used for detection is indicated at the bottom. After the blot was probed with anti-GFP (Upper), it was stripped and probed with anti-intein In (Lower).

The phenotype of the transplastomic plants was normal and the plants were fertile, although a high level of GFP and aadA accumulated in the cells. GFP expression levels were monitored by using a hand-held UV light. Control plants appeared red under UV illumination (data not shown). The whole transplastomic plant was green under both white (Fig. 3C) and UV (Fig. 3D) illumination, suggesting a significant masking of the red color of the endogenous chlorophyll by high levels of GFP. To demonstrate that GFP is produced only in the chloroplast, protoplasts from both control and transplastomic lines were isolated and examined under a fluorescent microscope with a UV light source. Chloroplasts from the control plant were bright red (Fig. 3E), whereas the transplastomic chloroplasts were green (Fig. 3F). We further examined the level of trans-splicing of aadA and smGFP in the transplastomic plants by Western blot. Cell extracts from the transplastomic plants showed aadA-smGFP fusions at 57 kDa (Fig. 3G), confirming protein trans-splicing in plant chloroplast. Only ≈20% of the precursor Ic-smGFP was converted to spliced product. To confirm trans-splicing in vivo (not in the plant extract) tissue samples were prepared in EB (without EDTA) containing 2 mM ZnCl₂ (Zn²⁺), a potent inhibitor of protein trans-splicing (15, 16). Western blot of the plant extract with anti-GFP showed aadA-smGFP trans-spliced product at 57 kDa. Intensity of this band in the presence or absence of Zn²⁺ was comparable. This result demonstrates that trans-splicing occurs in vivo and not in the cell extract where Zn²⁺ is present (Fig. 3H). The blot was stripped and probed with anti-In antibody to determine the extent of aadA-In precursor utilization. Because smGFP was codon-optimized for expression in plant tissue, Ic-smGFP expression was several times higher than aadA-In. Therefore, all of the aadA-In molecules were converted to aadA-smGFP product. Only In cleaved product of ≈14 kDa was observed (Fig. 3H Lower).

Translocalization of Nuclear-Encoded EPSPSn-In into Chloroplast and Trans-Splicing with Ic-EPSPSc.

Because protein trans-splicing was efficient in the chloroplast, we investigated whether such a system could support reconstruction of a full-length EPSPS. An EPSPS system was used in the current experimental design along with wild-type In and Ic intein fragments. The rationale was to introduce one part of the EPSPS gene into the nuclear genome and the second part into the chloroplast genome. After translation in the cytoplasm, the nuclear-encoded gene product will migrate to the chloroplast and trans-splice with the chloroplast-encoded peptide, making a full-length EPSPS (Fig. 4A). Thus, the S. typhimurium EPSPS gene was split into two fragments, one containing amino acid residues 6–235 and the other containing amino acid residues 236–427. The chloroplast localization signal peptide from petunia EPSPS was engineered at the N terminus of EPSPS fragment 6–235. The In gene was fused to the 3′ end of this DNA. This tripartite fusion protein (EPSPSn-In) should be capable of translocation into the chloroplast after translation in the cytoplasm with the help of the chloroplast localization signal. A. tumefaciens containing the EPSPSn-In cassette, pBIEPSPSn-In, was used for leaf disk transformation. A tripartite expressed protein was detected only in transgenic plant cells. The observed molecular mass of the tripartite fusion was ≈47 kDa, agreeing well with the calculated molecular mass (Fig. 4B). In a separate transformation, p226IcEPC (Ic-EPSPSc) was introduced into the chloroplast genome by means of the biolistic transformation approach. Transplastomic cell lines were examined for Ic-EPSPSc protein expression. Transgene expression levels in the transplastomic plants were much higher than nuclear transformed plants because of the copy number of the transgenes in the cells. A typical plant cell contains ≈1,000 chloroplasts (Fig. 4C). These results suggest that plant cells can support expression of the S. typhimurium EPSPS gene fragments from different organellar genomes. To study the trans-splicing event in the chloroplast, the nuclear-transformed plants with EPSPSn-In were transformed once again with p226IcEPC. Double-transformant plants were visible after 5–7 weeks. These plants were resistant to spectinomycin and kanamycin because of the presence of both of the marker genes. Mature plant leaves were harvested and checked for transgene integration by using Southern blot analysis and PCR (data not shown). The leaf extracts were resolved on an SDS-polyacrylamide gel, transferred to nitrocellulose membranes and probed with antibodies specific for the N or C terminus of S. typhimurium EPSPS. After double transformation, all N + C plants (transgenic plants expressing both N and C termini of EPSPS) displayed an ≈55-kDa band, the expected molecular mass of full-length EPSPS, after Western blot analysis with either EPSPS N (Fig. 4D) or C terminus (Fig. 4E Left) specific antibodies. In Fig. 4D, the precursor EPSPSn-In appears largely converted into spliced protein because of the excess amount of Ic-EPSPSc. An ≈10-fold excess of Ic-EPSPSc was observed in transgenic lines (Fig. 4E Right). The presence of full-length EPSPS demonstrates the occurrence of trans-splicing in the chloroplast. After 4 days of treatment with 0.2 mM glyphosate, the wild-type plants were bleached, whereas the trans-splicing lines remained green (Fig. 5). This result was similar to that of transgenic tobacco plant carrying an intact nuclear EPSPS gene (data not shown) and demonstrated that glyphosate resistance was conferred by the spliced EPSPS gene products.

Figure 4.

Trans-splicing of EPSPS in plant cells. (A) Diagram showing nuclear-encoded EPSPS-In (46 kDa) and chloroplast-encoded Ic-EPSPSc (26 kDa). After intein-mediated trans-splicing, the full-length EPSPS is ≈55 kDa. (B) Transgenic plant expressing EPSPSn-In. Plant lines are numbered on the top and the antibody used is at the bottom. (C) Another set of transplastomic plants expressing chloroplast-encoded Ic-EPSPSc fusion protein. Plant lines are indicated on the top and the antibody used is at the bottom. (D) Western blot analysis of reconstituted full-length EPSPS in transgenic plants with sequential transformation; EPSPSn-In integrated into the nuclear genome and Ic-EPSPSc integrated into the chloroplast genome. Plant lines are indicated on top. Antibody used was specific for EPSPSn. The extracts were separated on a 10% denaturing gel for blotting. (E) Identical to D except that the antibody for Western blot analysis was specific for EPSPSc. (Right) Both precursor and trans-spliced product. The extracts were separated on a 4–20% gradient denaturing gel for blotting. A specific band at 46 kDa with EPSPS C-terminal specific antibody is presumably a degradation product.

Figure 5.

Glyphosate resistance of 7(N + C) transgenic lines (Left) vs. wild-type tobacco plants (Right). Glyphosate concentration was 0.2 mM.

Discussion

Promiscuity in transgenic plants by pollen has been reported and cited as a potential environmental risk (17, 18). Genetically modified plants with a transgene integrated in the nucleus may be able to hybridize with sexually compatible species and give rise to hybrids and their progeny. The consequence of transgene transfer from genetically modified crops to wild relatives will give selective advantage to the new hybrid. We proposed that protein trans-splicing might be applied to prevent transgene flow to weedy relatives of commercially important crop species. Transgene containment by splitting the gene requires that both partial proteins are inactive and can come together to make a full-length protein in a predetermined organelle or tissue. In this work, we have demonstrated robust protein splicing in plant chloroplasts. Although protein splicing in plant chloroplasts is but one example, we clearly envision applicability to a variety of other selective factors, giving the prospect of genetically modified plants with greatly reduced risk of spreading the resistance factor throughout the environment (19).

In the specific example described herein we have placed a portion of the EPSPS gene in the nuclear genome with the remaining portion placed into the chloroplast genome. These plants were able to make full-length EPSPS in the chloroplast. Because chloroplasts are maternally inherited in most plant species (20), pollen-mediated transgene transfer will be limited to the gene half in the nucleus of those species, resulting in an inactive peptide fragment. Further containment could be accomplished by localizing the nuclear fragment on specific chromosomes of cultivated species that are incompatible with related weed genomes. Cultivated oilseed rape and wheat have multiple genomes acquired from different wild relatives. Often, only one genome of the crop is compatible for interspecific hybridization with that of a related weed, allowing gene transfer to the weed. For example, the B genome of oilseed rape (Brassica juncea, Indian or brown mustard) is capable of spreading transgenes to other Brassica species (21). Similarly cultivated wheat genomes (AABBDD) are derived from different wild relatives' genomes (22). Only the D genome is compatible for interspecific hybridization with the D genome of Aegilops cylindrica (bearded goat grass), a wild relative of cultivated wheat allowing transgene migration (23). Therefore, introducing a transgene fragment fused to the Ssp DnaE intein In into the A genome and the rest fused to Ic into the B genome of wheat and allowing the trans-splicing to occur in the cytoplasm would offer alternative protection against transgene migration. Furthermore, transgenic plants with a portion of the gene in the chloroplast and the rest in an incompatible genome would make the transgene migration virtually impossible. It is also possible that direct uptake of a whole transgene by soil microbes would be reduced because the gene fragments are located in two different genomes.

As yet, no single strategy has proved broadly applicable to all crop species for transgene containment, and a combination of approaches may prove most effective for environmentally safe transgenic crops. Thus, protein/enzyme trans-splicing in transgenic crops may play a significant role in transgene containment in transgenic crops.

Abbreviations

EPSPS

5-enolpyruvylshikimate-3-phosphate synthase

ALS

acetolactate synthase

smGFP

soluble modified GFP