We report plastid marker-gene excision with a transiently expressed CRE, site-specific recombinase. This is a novel protocol that enables rapid removal of marker genes from the approximately 10 000 plastid genome copies without transformation of the plant nucleus. Plastid marker excision was tested in tobacco plants transformed with a prototype polycistronic plastid vector, pPRV110L, designed to express multiple genes organized in an operon. The pMHB10 and pMHB11 constructs described here are dicistronic and encode genes for herbicide (bar) and spectinomycin (aadA) resistance. In vector pMHB11, expression of herbicide resistance is dependent on conversion of an ACG codon to an AUG translation initiation codon by mRNA editing, a safety feature that prevents translation of the mRNA in prokaryotes and in the plant nucleus. In the vectors, the marker gene (aadA) is flanked by 34-bp loxP sites for excision by CRE. Marker excision by a transiently expressed CRE involves introduction of CRE in transplastomic leaves by agro-infiltration, followed by plant regeneration. In tobacco transformed with vectors pMHB10 and pMHB11, Southern analysis and PCR identified approximately 10% of the regenerated plants as marker-free.
Because of readily obtainable high protein levels and natural containment, the plastid genome (ptDNA) is an attractive target for biotechnological applications. The approximately 150-kb ptDNA is highly polyploid and may be present in 1000 to 10 000 copies per cell. Plastid transformation involves targeted insertion of the transforming DNA into the plastid genome by two homologous recombination events (Figure 1a), and amplification of the rare, transformed copies by selection for antibiotic resistance encoded in the vector (Bock, 2001; Maliga, 2004). Once transformation of ptDNA is accomplished, the marker gene is no longer desirable because of the metabolic burden imposed by high levels of marker-gene product; and because of the requirement to use the same marker gene for multi-step engineering and to meet consumer expectations (Maliga, 2004).
There are multiple protocols for the excision of plastid marker genes. Most protocols are tedious because transformation and marker-gene elimination occur simultaneously (Iamtham and Day, 2000; Ye et al., 2003); or the protocols may depend on the availability of specialized plant lines (Klaus et al., 2004). Generally applicable, efficient protocols for marker-gene excision rely on a nuclear encoded, plastid-targeted CRE recombinase derived from the P1 bacteriophage, which excises marker genes flanked by directly oriented 34-bp loxP sites (Figure 1b) (Corneille et al., 2001; Hajdukiewicz et al., 2001). When removal of the plastid marker gene has been accomplished, the nuclear gene encoding CRE has to be removed by segregation in the seed progeny. Removal of the nuclear Cre is a time-consuming process and is not practical in vegetatively propagated crops, in which seed propagation is incompatible with variety preservation. We report here an alternative protocol relying on excision of the multi-copy plastid marker genes by a transiently expressed CRE recombinase in Nicotiana tabacum (tobacco).
Transplastomic tester plants
Vector pPRV110L is a member of the Plastid Repeat Vector (pPRV) family (Zoubenko et al., 1994), targeting insertions between the trnV and 3′-rps12 intergenic region (Shinozaki et al., 1986). In the vector the selectable spectinomycin-resistance (aadA) gene has no promoter only a ribosome-binding site (RBS) and the 3′-UTR of the plastid rbcL gene (TrbcL) for stabilization of mRNAs. Vector pPRV110L is a loxP vector, because loxP sites flank aadA so that the marker gene can be excised when the homoplastomic state is achieved (Figures 1b and 2a). Plastid transformation vectors pMHB10 and pMHB11 are dicistronic and encode the bar and aadA genes (Figure 2a), the expression of which confers herbicide resistance (Iamtham and Day, 2000; Lutz et al., 2001; Ye et al., 2003) and spectinomycin resistance (Goldschmidt-Clermont, 1991; Svab and Maliga, 1993), respectively. The pMHB10 and pMHB11 plasmids differ with respect to the leader sequence of the bar gene. The bar gene (b-bar2) N-terminus in plasmid pMHB10 is translationally fused with the PrrnLatpB + DS, the rrn operon promoter fused with the plastid atpB leader and 14 of the N-terminal atpB codons (Kuroda and Maliga, 2001). In vector pMHB11, the bar coding segment is translationally fused with the psbL gene N-terminus (Ebar gene). Translation of the tobacco plastid psbL gene is dependent on conversion of an ACG codon to an AUG translation initiation codon by mRNA editing (Kudla et al., 1992). The 73-nt segment (−63 to +10, plasmid pSC4) containing the psbL initiation codon, used in this study, was sufficient to direct efficient editing of the initiation codon of a chimeric transcript enabling expression of the encoded protein (Chaudhuri and Maliga, 1996). Therefore we expected that the chimeric pMHB11 Ebar mRNA would also be edited and herbicide resistance would be expressed.
Plasmids pMHB10 and pMHB11 were introduced into the plastid genome, and uniform transformation of plastid genomes was confirmed by DNA gel-blot analysis (Figure 2b). Three pMHB10 (Nt-pMHB10-13; Nt-pMHB10-14; Nt-pMHB10-23) and two pMHB11 (Nt-pMHB11-12; Nt-pMHB11-18) independently transformed lines were obtained. The transplastomes in these plants will be termed TP1-ptDNA (Figure 2a).
Editing-dependent bar is expressed in chloroplasts but not in Escherichia coli
The newly obtained transplastomic Nt-pMHB10 and Nt-pMHB11 plants were tested for herbicide resistance. The TP1-ptDNA leaf segments were resistant to 4 mg l−1 phosphinothricin in cell culture (Figure 3a). The regenerated plants were resistant to 2% Liberty herbicide, a commercial formulation of phosphinothricin, when sprayed in the glasshouse (Figure 3b). In accordance with the herbicide-resistance phenotype, phosphinothricin acetyl transferase (PAT) activity was detected in vitro by acetylation of phosphinothricin in the leaves of both Nt-pMHB10 and Nt-pMHB11 plants (Figure 4). Expression of Ebar in Nt-pMHB11 plants was enabled by post-transcriptional conversion of an ACG to an AUG translation initiation codon (Figure 5). Editing of the bar segment was efficient because most signal (>85%) is found in the T lane (edited nucleotide) instead of the C lane (non-edited nucleotide) (Figure 5).
We also tested PAT activity in E. coli carrying the plastid transformation vectors. PAT activity could be readily detected in E. coli cells carrying the pMHB10 construct, which does not require mRNA editing for expression; but not in cells carrying the pMHB11 construct (Figure 4), indicating that the Ebar gene is not edited and therefore is not expressed in E. coli.
Marker excision in agro-infiltrated leaves
The main question addressed here is whether or not transient expression yields sufficient CRE levels to excise a significant fraction of the marker genes from the approximately 10 000 TP1-ptDNA copies to enable recovery of marker-free (TP2-ptDNA) plants, as shown in Figure 1(b). Agro-infiltration is a protocol suitable for high-level nuclear gene expression from a transiently introduced T-DNA region. It involves vacuum treatment of leaf segments in an Agrobacterium suspension, and subsequent release of the vacuum to facilitate entry of bacterium cells into the intercellular space (Kapila et al., 1997). Protein expressed from the T-DNA accumulates 2–4 days after agro-infiltration (Johansen and Carrington, 2001; Voinnet et al., 2003). To test the efficiency of marker-gene excision in TP1-ptDNA, leaf segments of Nt-pMHB10 and Nt-pMHB11 were infiltrated with Agrobacterium strain EHA101 (Hood et al., 1986) containing the pKO31 binary vector encoding a plastid-targeted CRE (Corneille et al., 2003). We tested marker-gene excision in TP1-ptDNA leaf segments up to 4 days after agro-infiltration. Marker-gene excision was observed as early as 3 days after agro-infiltration. By day 4, a significant fraction (10–20%) of plastid genome copies lacked the marker gene (Figure 6a).
Marker-free plants from agro-infiltrated leaves
Encouraged by marker-gene excision from TP1-ptDNA in the short-term experiment, we attempted to obtain marker-free transplastomic plants from Agrobacterium-treated leaves of Nt-pMHB10-13A, Nt-pMHB10-23A, Nt-pMHB11-12A, Nt-pMHB11-12B and Nt-pMHB11-12C plants. Plant regeneration was carried out on a medium containing phosphinothricin (PPT) to select for the bar gene as a second round of purification and carbenicillin (500 mg l−1) to kill the Agrobacterium cells. Plants regenerated on PPT medium were named by a second letter and randomly screened for aadA excision. Southern blots identified marker-excised (TP2-ptDNA) plastid genomes in 19 of the 61 regenerated plants (11 of 27 Nt-pMHB10; eight of 34 Nt-pMHB11). Examples for marker excision by DNA and RNA gel-blot analysis are shown in Figure 6(b,c), respectively. The nuclear Cre gene was absent in seven clones (Nt-pMHB10-13AB; Nt-pMHB10-23AB; Nt-pMHB10-23AI; Nt-pMHB11-12AK; Nt-pMHB11-12BV; Nt-pMHB11-12BW; Nt-pMHB11-12CF), as determined by PCR analysis. One of these (Nt-pMHB11-12AK) carries the unselected gentamycin-resistance gene encoded in pKO31 (data not shown).
Overall, excision of aadA by a non-integrated nuclear Cre was observed in approximately 10% (six out of 61) of the regenerated plants. Examples for this category in Figure 6(b) are lines Nt-pMHB10-23AI and Nt-pMHB11-12CF. Although a small fraction of ptDNA copies still may contain aadA in blots shown in Figure 6(b), these copies are typically lost during plant development. As an example of the absence of TP1-ptDNA, the Nt-pMHB10-23AI seed progeny, is shown in Figure 6(d).
Polycistronic loxP expression vector
Plastid vector pPRV110L is the prototype of a polycistronic plastid vector with an excisable marker gene. The vector is polycistronic, because it was designed for expression of multiple genes organized in an operon. The operon, with its own promoter, is cloned upstream of aadA, which becomes the last reading frame of the enlarged operon (Figure 2a). Such polycistronic expression units are advantageous because they minimize the number of promoters and terminators, which are necessary for expression of multiple transgenes. Constructs described here (pMHB10, pMHB11) are dicistronic and encode genes for herbicide and spectinomycin resistances. LoxP sites flank the aadA marker gene in vector pPRV110L so that it can be excised when the homoplastomic state is achieved.
Expression of herbicide resistance is dependent on mRNA editing
Expression of PAT confers resistance to glufosinate herbicides when expressed in chloroplasts (Iamtham and Day, 2000; Lutz et al., 2001; Ye et al., 2003). The bar gene in vector pMHB10 has the same promoter as plasmid pKO18 (PrrnLatpB + DS), a monocistronic gene we studied earlier (Lutz et al., 2001). In vector pMHB11, the PAT coding segment is translationally fused with the psbL gene N-terminus. The −63 to +10 psbL fragment was sufficient to ensure mRNA editing of the chimeric bar mRNA (Figure 5) so that expression of the Ebar gene confers herbicide resistance in Nt-pMHB11 plants. Similar editing-dependent genes were described by fusing psbL editing segments with aadA and neo coding regions, which conferred spectinomycin and kanamycin resistances when the chimeric mRNA was edited in chloroplasts (Chaudhuri and Maliga, 1996; Chaudhuri et al., 1995). Translation of edited mRNAs in a chimeric context thus far has been reported only for the aadA and neo genes. Expression of PAT and the associated herbicide-resistance phenotype demonstrate the general applicability of editing-dependent genes.
We have described a synthetic, codon-optimized plastid s-bar gene, which is not expressed in E. coli due to the abundance of codons that are rarely used in E. coli (Lutz et al., 2001). The editing-dependent Ebar gene (pMHB11) reported here is not expressed in E. coli (Figure 4b), because prokaryotes lack the capacity for mRNA editing (Smith et al., 1997). Dependence of Ebar expression on editing creates an effective expression barrier when the gene is transferred to prokaryotes, a phenomenon shown to occur under laboratory conditions at a very low frequency (Kay et al., 2003; Tepfer et al., 2003). As no mRNA sequence is edited in the plant nucleus (Smith et al., 1997), Ebar mRNA will not confer herbicide resistance if the gene accidentally escapes to the plant nucleus (Huang et al., 2003; Stegemann et al., 2003). However, this Ebar will be expressed in the plastids of tobacco, and other higher plant species which share the capacity to edit the initiation codon of psbL mRNA (Chateigner-Boutin and Hanson, 2002; Hegeman et al., 2005; Kudla et al., 1992).
Excision of marker genes by transiently expressed CRE
As there are 1000 to 10 000 ptDNA copies in a tobacco cell (Bendich, 1987), the challenge we have been facing was whether or not transient CRE expression is sufficient for excision of plastid marker genes from most genome copies. We report here that transient CRE expression is sufficient to obtain marker-free tobacco plants in the absence of Cre integration in the plant nucleus. The frequency of plastid marker-free plants in this study (10%) is comparable with the approximately 12% excision frequency (13 out of 57) of a floxed plastid codA gene in an independent experiment (data not shown).
Although excision of marker genes from the plant nuclear genome is routine (Gilbertson, 2003; Hare and Chua, 2002; Ow, 2002), the system we describe here is for the excision of organellar DNA by a transiently expressed CRE. So far only constitutively expressed CRE has been used for the excision of plastid marker genes (Corneille et al., 2001, 2003; Hajdukiewicz et al., 2001; Lutz et al., 2004) and to probe plastid gene function (Kuroda and Maliga, 2003). Data presented here show that a short burst of CRE expression is sufficient to eliminate target sequences from all genome copies in the highly polyploid plastid genetic system. Plastid marker-gene excision by a transiently expressed CRE is advantageous because it saves time, as there is no need to remove the CRE integrated in the nucleus. Transient CRE expression limits the time during which CRE may come into contact with ptDNA, thereby reducing the probability of undesired ptDNA rearrangements, including deletions involving fortuitous ptDNA sequences that function as loxP-sites (Corneille et al., 2003; Hajdukiewicz et al., 2001) and repeat-mediated recombination events, which are increased in the presence of CRE (Corneille et al., 2001; Mlynarova et al., 2002). Transient expression of CRE for marker-gene removal now enables production of marker-free transplastomic plants in vegetatively propagated species such as potato, in which variety preservation is incompatible with seed propagation.
Plastid transformation vectors pPRV110L (GenBank accession no. DQ211347); pMHB10 (GenBank DQ211346); and pMHB11 (GenBank DQ211345) have been constructed by standard molecular biology protocols, and their maps are shown in Figure 2(a). Vector pPRV110L targets insertions between the trnV and 3′-rps12 intergenic region (Shinozaki et al., 1986) (insertion site no. 1, first digit of 110L) carrying a selectable spectinomycin-resistance (aadA) gene (Chinault et al., 1986; Svab and Maliga, 1993) (marker gene no. 1, second digit) without a promoter but with a ribosome-binding site (RBS) and the 3′-UTR of rbcL gene (TrbcL) (Shinozaki and Sugiura, 1982) (expression signals no. 0, third digit). Vector pPRV110L is a loxP (L) vector because loxP sites (Hoess et al., 1982; Van Duyne, 2001) flank aadA so that the marker gene can be excised when the homoplastomic state is achieved (Figures 1b and 2a). Plastid transformation vectors pMHB10 and pMHB11 are dicistronic and encode the bar (Thompson et al., 1987) and aadA (Chinault et al., 1986) genes. Plasmid pMHB10 (GenBank DQ211346) carries the b-bar2 gene in which the bar N-terminus is translationally fused with the PrrnLatpB + DS (Kuroda and Maliga, 2001). Vector pMHB11 (GenBank DQ211345) carries the Ebar coding segment in which bar is translationally fused with the psbL gene N-terminus, including 63 nucleotides upstream and 10 nucleotides downstream of the psbL translation initiation codon, as in plasmid pSC4 (Chaudhuri and Maliga, 1996).
Plastid transformation and identification of transplastomic clones was carried out as described previously (Svab and Maliga, 1993). Briefly, sterile leaves were bombarded with DNA-coated tungsten particles using the biolistic gene gun (PDS-1000; Bio-Rad, Hercules, CA, USA). Plastid transformants were identified on RMOP medium containing 500 mg l−1 spectinomycin dihidrochloride by their green color and shoot formation, and uniform transformation of ptDNA genome copies was verified by DNA gel-blot analysis. Plants were designated by the plasmid name, a serial number, and letters to indicate the number of regeneration cycles on a selective medium to segregate away non-transformed ptDNA. An example of line designation is Nt-pMHB10-23AH, which involved two cycles of plant regeneration on a selective medium.
Vacuum infiltration was performed as described by Kapila et al. (1997). Agrobacterium transformed with pKO31 (Corneille et al., 2003), carrying the plastid-targeted cre gene, was inoculated into 100 ml yeast extract broth (5 g l−1 beef extract, 1 g l−1 yeast extract, 5 g l−1 peptone, 5 g l−1 sucrose, 2 mm MgSO4) supplemented with 100 mg l−1 spectinomycin, and grown overnight at 27°C. Culture (1 ml) was inoculated into fresh YEB containing 10 mm MES [2-(N-morpholino)ethanesulfonic acid], pH-adjusted to 5.6, 20 mm acetosyringone and 100 mg l−1 spectinomycin, and grown overnight at 27°C. The culture was centrifuged at 4000 g for 15 min, resuspended in MMA medium (MS salts, 10 mm MES, 20 g l−1 sucrose pH 5.6, 200 mm acetosyringone) to a final OD660 = 2.4 and incubated at 24°C for 1 h. Nt-pMHB10 and Nt-pMHB11 leaves were cut into small pieces (1 cm2) and submerged in the Agrobacterium culture. The flask containing culture was placed under continuous vacuum of 2 Torr for 20 min, while shaking gently. After vacuum infiltration, leaf samples were incubated on RMOP medium for 2 days, then transferred to RMOP medium containing carbenicillin (500 mg l−1). DNA was isolated daily from individual leaf samples for 4 days after vacuum infiltration. Southern blots were performed on total leaf cellular DNA samples from individual leaf pieces digested with the ApaI and EcoRI restriction enzymes and probed with the plastid targeting region (1.9 kb ApaI–StuI ptDNA fragment containing the rrn16 gene; Figure 2a) (Svab and Maliga, 1993). The remaining leaf pieces were incubated further on RMOP shoot-regeneration medium containing carbenicillin (500 mg l−1) to kill Agrobacterium cells. After 4 weeks, individual shoots were removed from the cultures and rooted on RM medium containing carbenicillin (500 mg l−1). Total cellular leaf DNA was probed for marker-gene (aadA) excision as described above.
Testing mRNA editing
Isolation of total cellular RNA and testing of RNA editing were performed by direct sequencing of reverse-transcribed, PCR-amplified (RT–PCR) cDNA, as described by Bock (1998). For RNA isolation, leaf tissue was ground in a mortar with a pestle and extracted with TRIzol (Invitrogen, Carlsbad, CA, USA). Total cellular RNA was treated with Proteinase K (1×, RNA grade, Invitrogen) and with DNase I (2×, Roche, Indianapolis, IN, USA) as described by Kudla et al. (1992). The absence of DNA contamination was confirmed by the lack of a PCR product in the control tube, which lacked reverse transcriptase. Complementary DNA was prepared with AMV (avian myeloblastosis virus) reverse transcriptase (GE Healthcare, Piscataway, NJ, USA) using random hexanucleotide primers. The cDNA samples were PCR-amplified using 1 μg each of primers 5′-GCGAATACGAAGCGCTTGG-3′ and 5′-GGCGACCTCGCCGTCCAC-3′, and one unit Taq polymerase (Applied Biosystems, Foster City, CA, USA). Amplification was performed as follows: 3 min at 92°C, followed by 28 cycles of 1 min at 92°C, 1 min at 55°C and 1.5 min at 72°C; 1 cycle of 1 min at 92°C, 1 min at 55°C and 11 min at 72°C; and 1 min at 30°C. PCR-amplification products were sequenced directly with the T7 sequenase PCR product-sequencing kit (United States Biochemical Co., Cleveland, OH, USA) using the above primers.
The phosphinothricin acetyl transferase (PAT) assay was performed as described by Spencer et al. (1990). Leaf tissue (100 mg) was homogenized in 1 vol extraction buffer (10 mm Na2HPO4, 10 mm NaCl). The supernatant was collected after spinning in a microfuge (14 000 g) for 10 min. Escherichia coli was grown to stationary phase levels (OD550 > 1.3). Lysate (400 μl) was collected and pelleted. The pellet was resuspended in 300 μl BugBuster protein extraction reagent (Novagen, Madison, WI, USA), incubated for 15 min on a shaker at 24°C and microfuged for 20 min. The Bio-Rad Protein Assay reagent kit was used to determine protein concentrations with bovine serum albumin as a reference; PAT activity in leaf extracts was determined using 10 μg protein. Protein extracts from bacteria were diluted 10-fold (2.5 or 0.25 μg protein per assay). PPT (1 mg ml−1) and 14C-labeled acetyl CoA were added to the protein samples and incubated at 37°C for 30 min; the entire reaction was spotted onto a TLC plate. Ascending chromatography was performed in a 3:2 mixture of 1-propanol and NH4OH, and radioactivity was detected by exposure to Kodak XAR6 film.
RNA gel-blot analysis
Total cellular RNA was prepared from the leaves of plants grown in sterile culture (Stiekema et al., 1988). The RNA (5 μg per lane) was electrophoresed on 1.2% agarose/formaldehyde gels, and then transferred to Hybond N membranes (GE Healthcare) using the PosiBlot Transfer apparatus (Stratagene, La Jolla, CA, USA). Hybridization with the bar probe was carried out in rapid hybridization buffer (GE Healthcare) overnight at 65°C. A double-stranded DNA probe was prepared by random-primed 32P-labeling. The template for probing bar was a gel-purified NheI–BglII fragment excised from plasmid pMHB10.
This research was supported by grants from the National Science Foundation Eukaryotic Genetics Program MCB-0319958 and a Rutgers Facilities and Administration Special Research Grant. K.A.L. was the recipient of a Charles and Johanna Busch Predoctoral Fellowship.
Accession numbers: Ebar gene in plasmid pMHB11, DQ211345; b-bar2 gene in plasmid pMHB10, DQ211346; vector pPRV110L, DQ211347.