Construction and Evaluation of Chloroplast Expression Vectors in Higher Plants

Plastid transformation has a number of advantages in comparison with nuclear transformation. Currently, only tobacco ( Nicotiana tabacum ) is routinely used in plastid transformation. Here we constructed a series of chloroplast expression vectors specific for spinach (pCEV1), tomato (pCEV2 and pCEV3), and N. benthamiana (pCEV4). Selection marker aminoglycoside 3 0 -adenyltransferase ( aadA ) conferring spectinomycin resistance was used in pCEV1, pCEV2, and pCEV4, while selection marker neomycin phosphotransferase II ( nptII ) was used in pCEV3. The expression cassette in these vectors was integrated in the intergenic spacer between trnI and trnA of plastid genome via homologous recombination. Several transgenes, including a reporter gene encoding GFP:GUS fusion protein and genes from tomato (lycopene b-cyclase, z-carotene desaturase) and bamboo mosaic virus satellite RNA (encoding coat protein CP20), were independently cloned into some of these vectors. Transient GUS expression was detected in spinach leaves bombarded by pCEV1/GFP-GUS. Functional expression of selection markers aadA and nptII was demonstrated for spinach, tomato, and N. benthamiana . Seedling assay from T 0 self-pollinated plant of transplastomic N. benthamiana confirmed maternal inheritance of transgenes, and genomic PCR analysis confirmed integration of transgenic expression cassette into the plastid genome of N. benthamiana . Moreover, auxiliary vectors pECaad and pECnpt are also reported.


Introduction
Plastid transformation has a number of advantages in comparison with nuclear transformation. It has high levels of transgene expression, position effect or gene silencing is not found due to precise integration of the transgene into the host plastid genome via homologous recombination, and further, gene flow through pollen is not an issue due to maternal inheritance. Plastid transformation technology is important for analyzing the functions of chloroplast genes and has also been widely used in mass production of foreign proteins for agronomic traits (such as insect resistance, herbicide resistance, disease resistance, drought resistance, salt resistance, phytoremediation, etc.), enzymes/biomaterials (such as cellulase, endo-glucanase, monellin, polyhydroxybutyrate, xylanase, β-glucosidase, pectin lyase, manganese peroxidase, superoxide dismutase, etc.), and pharmaceuticals/ vaccine antigens (such as interferon-γ, human serum albumin, human somatotropin, insulin-like growth factor, thioredoxin 1, basic fibroblast growth factor, aprotinin, cholera toxin, tetanus toxin, Canine parvovirus, anthrax protective antigen, dengue virus, etc.) over the past few decades [1][2][3][4][5][6]. In most cases, only tobacco (Nicotiana tabacum) is routinely chosen as a target plant in plastid transformation [1]. Besides tobacco, a few studies using tomato, lettuce, oilseed rape, potato, cabbage, cotton, carrot, N. benthamiana, etc., as target plants have also been reported to produce foreign proteins by plastid transformation technology [1][2][3][4][5][7][8][9][10].
In nuclear transformation, one nuclear transformation vector can be applied to multiple plants. For example, a nuclear transformation vector pCHS [carrying the expression cassette for the selection marker neomycin phosphotransferase II (nptII) driven by nopaline synthase (nos) promoter and Petunia chalcone synthase (chs) driven by cauliflower mosaic virus 35S promoter] and an Agrobacterium tumefaciensmediated method were employed to a range of plant species including the highvalue medicinal plants Echinacea purpurea [11] and E. pallida [12], model plant tobacco [13], medicinal herb Bidens pilosa [14], and the floricultural plant Cleome spinosa [15]; the transfer DNA (T-DNA) containing the expression cassette in vector pCHS was integrated randomly in the nuclear chromosome of transgenic plants, and the foreign Petunia chs and selection marker nptII were detected and expressed. However, in plastid transformation, a plastid transformation vector has to be constructed for each target plant which is quite a laborious work. A typical plastid transformation vector consists of two components: (1) an expression cassette [a gene of interest (GOI) which is inserted between the plastid promoter and the plastid terminator, followed by a selection marker gene which is inserted between the plastid promoter and the plastid terminator] and (2) a targeting sequence for homologous recombination in the host plant plastid genome. The expression cassette is located between the left targeting region (LTR) and the right targeting region (RTR) in the vector, and this expression cassette is expected to integrate into the plastid genome of the host plant through double homologous recombination during plastid transformation [1,[3][4][5]. Thus, in comparison with random insertion in nuclear transformation, the expression cassette is precisely inserted at the predicted integration site of the plastid genome during plastid transformation. In theory, any plastid DNA fragment within the plant plastid genome can serve as a targeting region in plastid transformation, and the integration site for the foreign expression cassette into the plant plastid genome can be anywhere within the targeting region. In practice, only 10 or so plastid DNA sequences have been chosen as targeting regions in plastid transformation [3]. Among them, the plastid trnI/ trnA sequence is the most frequently used as a targeting region, and the spacer between trnI and trnA is the most frequently used as the integration site [1,3]. In addition, the selection marker aminoglycoside 3 0 -adenyltransferase (aadA) conferring resistance to spectinomycin or streptomycin is most frequently used in plastid transformation, although other selection markers such as antibiotic resistance (including nptII), herbicide resistance, photosynthesis, and metabolism have been reported [16]. More recently, a new selection marker, namely, aminoglycoside acetyltransferase/phosphotransferase, conferring tobramycin in tobacco plastid transformation has been developed [17]. Spinach (Spinacia oleracea), belonging to Chenopodiacea, is a leaf vegetable which is an important source of vitamins and other essential nutrients in many countries. Callus induction and plant regeneration from different explants including leaves, cotyledons, and hypocotyls of several commercial spinach cultivars have been reported [18,19]. For nuclear transformation, an Agrobacterium-mediated method was reported from leaf and hypocotyl explants of spinach cultivars Fall Green and High Pack, and transgenic spinach plants expressing cucumber mosaic virus coat protein were obtained [20]. In addition, Agrobacterium-mediated transformation was also established from cotyledon explants of spinach cultivars Longstanding Bloomsdale and Melody, and transgenic spinach plants expressing reporter gene encoding green fluorescent protein (GFP) were regenerated [21]. However with regard to plastid transformation, although the complete chloroplast genome of spinach has been published [22], plastid transformation in spinach has not yet been reported. To initiate plastid transformation in spinach, we constructed chloroplast expression vectors specific for spinach. During the construction, we found that we can easily replace the host plant targeting sequence, which is crucial for homologous recombination in plastid transformation, from spinach to other plant species. Tomato (Solanum lycopersicum) was chosen as another target plant in this study, since tomato is an important foodstuff worldwide. Although plastid transformation was reported in tomato cultivar Santa Clara [7], we still wanted to set up plastid transformation technology for tomato cultivar CL5915, a local Taiwan cultivar. N. benthamiana was the third target plant in this study. Similar to the widely used tobacco (N. tabacum), N. benthamiana has many advantages in cell culture such as being easy to regenerate, short lifetime, and self-pollination. Two more advantages in comparison with N. tabacum are the plant height and leaf size which are much reduced in N. benthamiana, allowing it to serve as model plant for fundamental research such as studies of host/virus interactions.
We want to set up chloroplast transformation technology in our laboratory. As a first step in achieving the goal, we need to construct various chloroplast expression vectors for different plant species. Here we describe the efficient construction of a series of chloroplast expression vectors specific for spinach (pCEV1), tomato (pCEV2 and pCEV3), and N. benthamiana (pCEV4). Selection marker aadA conferring spectinomycin resistance was used in pCEV1, pCEV2, and pCEV4, while selection marker nptII conferring kanamycin resistance was used in pCEV3. Subsequently, several transgenes, including reporter gene encoding GFP:GUS fusion protein and genes from tomato [lycopene β-cyclase (LCY), ζ-carotene desaturase (ZDS)] and bamboo mosaic virus satellite RNA (encoding coat protein CP20), were independently cloned into some of these vectors. The reporter gene encoding GFP: GUS fusion protein was inserted into all plastid vectors, so that the expression of the transgene can be visualized by GFP fluorescence microscopy or GUS staining. LCY and ZDS are key enzymes in the carotenoid biosynthesis pathway. Previously, a few genes including LCY and ZDS have been cloned from tomato in our laboratory [23]. In this study, LCY and ZDS were selectively inserted into pCEV1 to investigate whether we can engineer spinach by using plastid transformation technology. CP20, encoded by bamboo mosaic virus satellite RNA, is the RNA-binding protein essential for virus infection of the host plant [24]. Thus, the coding region of CP20 with or without untranslated regions was inserted into pCEV4 for N. benthamiana. Accuracy of cloning GOI into these vectors has been demonstrated, and functional expressions of reporter and selection marker genes have also been confirmed. In addition, we further report two auxiliary vectors pECaad (carrying selection marker aadA) and pECnpt (carrying selection marker nptII) which may facilitate vector construction in other plant species.

Plant material and culture conditions
Seeds of spinach (Spinacia oleracea), tomato (Solanum lycopersicum var. CL5915), tobacco (Nicotiana tabacum var. W38), and N. benthamiana were individually sterilized by sequential treatment with 70% ethanol for 0.5 min and 1% sodium hypochlorite for 10 min, then washed thoroughly with sterile water, and germinated on MS basal medium (MS salts [25]; 3% sucrose; 0.8% Bacto agar; pH 5.7). The cultures were then incubated in a growth chamber at 22°C under a photoperiod of 16 h illumination (100 μmol m À2 s À1 ) and 8 h darkness. In vitro explants were used to establish the plant regeneration and plastid transformation.

Construction of chloroplast expression vector pCEV1
Schematic representation of the cloning procedure used to obtain a chloroplast expression vector in this study is shown in Figure 1, and the list of the relevant primers used in this study is shown in Table A1. Genomic DNA from green leaves of the target plants was isolated by the cetyltrimethyammonium bromide (CTAB) method [26]. DNA fragments of Prrn, TpsbA, PatpI, and Trps16 were amplified in 50 μl PCR mixture [1Â PCR buffer; 200 μM dNPT; 0.5 μM each of gene-specific forward and reverse primers; 500 ng tobacco (N. tabacum) genomic DNA; 2.5 U FastTaq DNA polymerase]. The PCR mixture was denatured at 94°C for 5 min prior to 35 amplification cycles (1 min at 94°C, 1 min at 55°C, 1 min at 72°C). Final extension reaction was performed at 72°C for 10 min. Following amplification, PCR products were analyzed on a 1% agarose gel, and PCR fragments were cut out of the gel, purified by the Zymoclean gel DNA recovery kit (Zymo Research, Orange, CA, USA), cloned into pGEM-T Easy vector (3015 bp; Promega), and then sequenced by an automatic DNA sequencer (ABI PRISM 377 DNA Sequencer, PerkinElmer).
Stepwise protocols for cloning pCEV1 (Figure 2) were as follows.

Cloning of Prrn-aadA-TpsbA fragment into pUC19
a. TpsbA (222 bp) was inserted into pUC19: As mentioned above, PCR amplification was carried out using TpsbA-F (containing XbaI site) and TpsbA-R (containing PstI site) as primers and 500 ng tobacco genomic DNA as template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector. c. aadA (792 bp) was inserted between Prrn and TpsbA in pUC19: PCR amplification was carried out using aadA-F (containing the BamHI site) and aadA-R (containing the XbaI site) as primers and 500 ng plasmid DNA of Step-by-step cloning protocol. PCR amplification was carried out using 16S/trnI-F (containing the SacI site) and 16S/trnI-R (containing the kpnI site) as primers and 500 ng of spinach (S. oleracea) genomic DNA as DNA template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector. Plasmid carrying SacI-LTR-KpnI was isolated and double digested by SacI and KpnI and then ligated with pUC19/Prrn-aadA-TpsbA which was double digested by SacI and KpnI, resulting in pUC19/LTR + (Prrn-aadA-TpsbA).

Cloning of spinach right targeting region (RTR; 2031 bp fragment containing fulllength trnA and partial 23S rRNA in spinach)
PCR amplification was carried out using trnA/23S-F (containing the PstI site) and trnA/23S-R (containing the PstI site) as primers and 500 ng of spinach genomic DNA as DNA template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector. Plasmid carrying PstI-RTR-PstI was isolated and digested by PstI and then ligated with pUC19/LTR + (Prrn-aadA-TpsbA) which was digested by PstI, resulting in pUC19/LTR + (Prrn-aadA-TpsbA) + RTR.

Cloning of PatpI-(cloning sites)-Trps16 into pUC19/LTR + (Prrn-aadA-
TpsbA) + RTR a. Preparation of PatpI (248 bp) fragment: PCR amplification was carried out using PatpI-F (containing the KpnI site) and PatpI-R (containing the NotI and SacII sites) as primers and 500 ng tobacco genomic DNA as template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector.
b. Preparation of Trps16 (169 bp) fragment: PCR amplification was carried out using Trps16-F (containing the SacII and AscI sites) and Trps16-R (containing the KpnI site) as primers and 500 ng tobacco genomic DNA as template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector.

Cloning of GOI into pCEV1
PCR amplification was carried out using LCY-F (containing NotI site) and LCY-R (containing AscI site) as primers and 500 ng plasmid DNA of our tomato cDNA clone encoding lycopene β-cylase (LCY, 1503 bp; Accession no. EF650013) as DNA template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector. Plasmid carrying NotI-LCY-AscI was isolated and double digested by NotI and AscI and then ligated with pCEV1 which was double digested by NotI and AscI, resulting pCEV1/LCY (9878 bp; Figure 3b). Similarly, PCR amplification was carried out using ZDS-F (containing NotI site) and ZDS-R (containing AscI site) as primers and 500 ng plasmid DNA of our tomato cDNA clone encoding ζ-carotene desaturase (ZDS, 1767 bp; Accession no. EF650012) as DNA template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector. Plasmid carrying NotI-ZDS-AscI was isolated and double digested by NotI and AscI and then ligated with pCEV1 which was double digested by NotI and AscI, resulting in pCEV1/ZDS (10,142 bp; Figure 3c). For cloning the gene (2568 bp) encoding mGFP5-GUSA fusion protein, PCR amplification was carried out using mGPF5-F (containing NotI site) and GUSA-R (containing AscI site) as primers and 500 ng plasmid DNA of pCAMBIA1304 (Center for the Application of Molecular Biology to International Agriculture, Australia) as DNA template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector. Plasmid carrying NotI-GFP-GUS-AscI was isolated and double digested by NotI and AscI and then ligated with pCEV1 which was double digested by NotI and AscI, resulting in pCEV1/GFP-GUS (10,943 bp; Figure 3d).

Cloning of GOI into pCEV2
As described in pCEV1, for cloning the gene (2568 bp) encoding GFP:GUS fusion protein, PCR amplification was carried out using mGPF5-F (containing NotI site) and GUSA-R (containing AscI site) as primers and 500 ng plasmid DNA of pCAMBIA1304 as DNA template. After PCR amplification, PCR products were purified and cloned into pGEM-T Easy vector. Plasmid carrying NotI-GFP-GUS-AscI was isolated and double digested by NotI and AscI and then ligated with pCEV2 which was double digested by NotI and AscI, resulting in pCEV2/GFP-GUS (10,921 bp).
For cloning of GOI into pCEV4, PCR amplification was carried out using NotI-BaMV_S-F (containing the NotI site) and BaMV_S_CP-AscI-R (containing the AscI site) as primers and 500 ng plasmid pBS2-8 DNA carrying bamboo mosaic virus strain S complete genome (6366 bp; GenBank Accession number AF018156) as DNA template. After PCR amplification, PCR products (753 bp containing enzyme cutting sites) for coding region of CP20 were purified and cloned into pGEM-T Easy vector. Plasmid carrying NotI-BaMV_S_CP-AscI was isolated and double digested by NotI and AscI and then ligated with pCEV4 which was double digested by NotI and AscI, resulting in pCEV4/BaMV (8975 bp). In addition, PCR amplification was carried out using NotI-BSL6-F (containing the NotI site) and BSL6-AscI-R (containing the AscI site) as primers and 500 ng plasmid pBSL6 DNA [carrying 5 0untranslated region (5 0 -UTR), coding region of CP20, and 3 0 -UTR] as DNA template. After PCR amplification, PCR products (861 bp containing enzyme cutting sites) of BSL6 (i.e., CP20 containing 5 0 -UTR and 3 0 -UTR) were purified and cloned into pGEM-T Easy vector. Plasmid carrying NotI-BSL6-AscI was isolated and double digested by NotI and AscI and then ligated with pCEV4 which was double digested by NotI and AscI, resulting in pCEV4/BSL6 (9083 bp). For cloning the gene encoding mGFP5-GUSA fusion protein, the same strategy was used as for pCEV1. In brief, PCR products of NotI-GFP-GUS-AscI were isolated and double digested by NotI and AscI and then ligated with pCEV4 which was double digested by NotI and AscI, resulting in pCEV4/GFP-GUS (10,814 bp).

Bombardment and transient GUS expression in spinach leaves
Spinach (S. oleracea cv. Green Giant; Known-You Seed Company, Taiwan) young leaves were excised from in vitro-grown plantlets and inoculated on culture medium [MS salts and vitamins (Duchefa Biochemie B.V., The Netherlands); 3% sucrose; 0.4 mg l À1 NAA; 1.0 mg l À1 BA; 1.0 mg l À1 GA 3 ; 0.8% Bacto agar, pH 5.8] in the dark for 3 days. Then, leaves were bombarded by the biolistic bombardment device (PDS-1000/He Particle Delivery System, Bio-Rad) with 1.25 mg of 0.6 μm gold particles (Bio-Rad) coated with 5 μg of plasmid DNA (pCEV1/GFP-GUS or pCAMBIA1305.1) using 1100 psi rupture disks (Bio-Rad). The distance from the uncovered Petri dish containing the leaf sample to the top of the PDS chamber was set to 9 cm, and the vacuum was set to 27 mm Hg. After bombardment, the samples were kept in the dark for 1 day.

Plastid transformation in spinach and tomato
Leaf segments were cut from in vitro-grown plantlets of spinach and inoculated on SO culture medium (MS salts and vitamins; 3% sucrose; 0.5 mg l À1 2,4-D; 2 mg l À1 kinetin; 0.8% Bacto agar; pH 5.8) for 1 day in a growth chamber with a cycle of 8 h light at 22°C and 16 h dark at 20°C. Conditions for bombardment were as described above, except that plasmid pCEV1/GFP-GUS, pCEV1/LCY, or pCEV1/ ZDS was used in each individual experiment. After bombardment, the sample was cultured in the dark for 4 days. Then, the bombarded leaf segments were trimmed into small pieces of 2-4 mm 2 , transferred onto SO selection medium (SO culture medium supplemented with 300 mg l À1 spectinomycin), and cultured for 1 month in a growth chamber with a cycle of 8 h light at 22°C and 16 h dark at 20°C. Survival green calli were picked up and transferred onto fresh SO selection medium and cultured for another 1 month in the same growth chamber.
For plastid transformation in tomato, leaf segments were cut from plantlets of tomato (S. lycopersicum cv. CL5915) grown in vitro and inoculated on SL culture medium (MS salts and vitamins; 3% sucrose; 0.2 mg l À1 2,4-D; 0.1 mg l À1 kinetin; 0.8% Bacto agar; pH 5.8) for 1 day in a growth chamber with a cycle of 8 h light at 22°C and 16 h dark at 20°C. Conditions for bombardment were as described above, except that plasmid pCEV2/GFP-GUS or pCEV3/GFP-GUS was used in each individual experiment. After bombardment, the sample was cultured in the dark for 3 days. Then, the bombarded leaf segments were trimmed into small pieces of 2-4 mm 2 , transferred onto SL selection medium (MS salts and vitamins; 3% sucrose; 1 mg l À1 NAA; 1 mg l À1 zeatin; 0.8% Bacto agar; pH 5.8), and supplemented with 300 mg l À1 spectinomycin (for pCEV2/GFP-GUS) or 50 mg l À1 kanamycin (for pCEV3/GFP-GUS) for 1 month in the same growth chamber. Survival green calli were picked up and transferred on fresh selection medium for another 1 month.

Plastid transformation and selection of transplastomic lines in N. benthamiana
Leaf segments were cut from in vitro-grown plantlets of N. benthamiana and inoculated on NB culture medium (MS salts and vitamins; 3% sucrose; 0.1 mg l À1 NAA; 1.0 mg l À1 BA; 0.8% Bacto agar; pH 5.8) for 1 day in a growth chamber with a cycle of 8 h light at 22°C and 16 h dark at 20°C. Conditions for bombardment were previously described, except that plasmid pCEV4/BaMV or pCEV4/BSL6 was used in each individual experiment. After bombardment, the sample was cultured in the dark for 3 days. Then, the bombarded leaf segments were trimmed into small pieces of 2-4 mm 2 , transferred onto NB selection medium (NB culture medium supplemented with 300 mg l À1 spectinomycin), and cultured for 1 month in a growth chamber with a cycle of 8 h light at 22°C and 16 h dark at 20°C. Survival green shoots were excised and transferred onto MS basal medium for rooting. The rooted plantlets were transferred into pots and grown in a greenhouse.

Transplastomic plant verification by PCR analysis
Transplastomic plant verification was carried out with putative transformants and wild-type N. benthamiana plants by PCR analysis. Total genomic DNA was extracted from green leaves of putative transformants and wild-type plants using the CTAB method [26]. PCR was carried out with the following sets of primers (Table A1): BSL-F and BSL-R for amplification of 837-bp-long DNA fragment corresponding to the 5 0 UTR coding region and 3 0 UTR of coat protein (CP20) in BaMV strain S (GenBank Accession no. AF018156), BaMV-F and BaMV-R for amplification of 729-bp-long DNA fragment corresponding to the coding region of CP20 in BaMV strain S, TpsbA-F and trnI-trnA-r for amplification of 351-bp-long DNA fragment corresponding to the truncated sequence composed of the transgenic TpsbA and partial RTR of N. benthamiana chloroplast genome, and trnA-f and 23S-rRNA-r for amplification of 342-bp-long DNA fragment located within the RTR of N. benthamiana chloroplast genome. PCR reactions were performed under the following conditions: (a) denaturation for 5 min at 94°C; (b) 35 cycles of 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C; and (c) extension for 10 min at 72°C. Following amplification, PCR products were analyzed by electrophoresis.

2.10
Seedling assay for spectinomycin resistance T 1 seeds from a self-pollinated transplastomic plant were sterilized in 70% ethanol for 0.5 min followed by 1% sodium hypochlorite for 10 min and washed thoroughly with sterile distilled water. The sterile seeds were germinated on a medium containing MS basal medium supplemented with 300 mg l À1 spectinomycin in a 25°C growth chamber under 16 h illumination for 4 weeks. Seedlings with green cotyledons were considered to be resistant, while those with white cotyledons were sensitive.

Construction of spinach chloroplast expression vectors
Although the complete chloroplast genome of spinach has been published [22], plastid transformation in spinach has not yet been reported. As a first step in achieving plastid transformation in spinach, spinach-specific chloroplast expression vectors must be constructed. As shown in Figure 1, the integration site for the expression cassette is designed at the spacer region between trnI and trnA. This integration site at trnI/trnA has been commonly used in constructing chloroplast expression vectors in different plants [1]. In addition, the trnI gene contains the chloroplast replication origin which might promote replication of foreign vectors within chloroplasts and enhance the transgene integration [1,29]. Another feature of our chloroplast expression vectors is the use of longer sequences for targeting via homologous recombination: the left targeting region (ca. 2 kb) contains partial 16S rRNA and full-length trnI, whereas the right targeting region (ca. 2 kb) contains full-length trnA and partial 23S rRNA (Figure 1). Targeting sequences are usually 1 kb each in size and are located on both sides of the expression cassette [1]. However, a 4 kb DNA fragment containing 16S rRNA-trnI/trnA-23S rRNA (i.e., 2 kb each on either side of the expression cassette), which is twice the size of the flanking region commonly used in chloroplast transformation, was used in cotton chloroplast transformation [9]. It has been reported that the frequency of homologous recombination is linearly reliant on the length of the homologous sequences [30]. Thus, in this study, each of the left/right targeting sequences was increased to 2 kb (total 4 kb in length), similar in size to that described in cotton chloroplast transformation [9].
Next, each DNA fragment including the left and right targeting regions, synthetic unique cloning sites (NotI, SacII, AscI) flanked by atpI promoter and rps16 terminator, as well as aadA selection marker gene flanked by rrn promoter and psbA terminator (Figure 1) were amplified by PCR and cloned and inserted into pUC19 according to a step-by-step protocol (Figure 2), and finally a chloroplast expression vector for spinach (designated as pCEV1) was constructed (Figure 3a). Any GOI may be cloned into pCEV1 by using the restriction enzyme NotI, SacII, or AscI. However in this study, pCEV1 was double digested by NotI and AscI and ligated with a PCR fragment of the GOI carrying NotI or AscI restriction site at each end. As a result, pCEV1/LCY [carrying full-length cDNA (1503 bp) encoding LCY from tomato], pCEV1/ZDS [carrying full-length cDNA (1767 bp) encoding ZDS from tomato], and pCEV1/GFP-GUS [carrying mGFP5-GUSA fusion protein gene (2568 bp) from pCAMBIA1304] were constructed (Figure 3). To check the accuracy of the cloning sites, plasmid DNA was isolated from E. coli carrying pCEV1, pCEV1/ZDS, pCEV1/LCY, or pCEV1/GFP-GUS and double digested with NotI and AscI. As shown in Figure 4, only a large fragment (it should be 8367 bp) was detected in pCEV1 after double digestion was carried out; the small fragment (it should be 8 bp) was not observed. As predicted, the same size large fragment (it should be 8367 bp) was detected in pCEV1/ZDS, pCEV1/LCY, or pCEV1/GFP-GUS; however, different sizes of small fragment including coding regions of LCY (1503 bp), ZDS (1767 bp), and GFP-GUS (2568 bp) were easy to detect and match according to the prediction (Figure 4). Thus, we concluded that spinach chloroplast expression vector pCEV1 and its derivatives were successfully constructed.

Transient GUS expression and development of plastid transformation in spinach
To evaluate the functionality of the constructed vector, spinach chloroplast expression vector pCEV1/GFP-GUS [carrying a mGFP5-GUSA fusion protein gene driven by plastid atpI promoter (PatpI) from tobacco] (Figure 3d) or positive nuclear control pCAMBIA1305.1 [carrying GUSPlus driven by cauliflower mosaic virus (CaMV) 35S promoter] was used to independently bombard the spinach leaves followed by GUS staining. As shown in Figure 5d, no blue spots were detected in leaf samples without bombardment (negative control). Only one to two very small blue spots per leaf were detected in leaves bombarded with pCEV1/GFP-GUS (Figure 5a and b). This is because to display blue spots, the golden particles carrying pCEV1/GFP-GUS need to pass through the cell wall and then the cell membrane and then the chloroplast membrane into the chloroplast of a cell, for which the probability is small, so the tiny but condensed blue spots indicated that GUS expression was indeed detected within chloroplasts. On the other hand, more than 10 blue spots per leaf were detected when bombardment was carried out with pCAMBIA1305.1, and the sizes of the blue spots were much larger than those blue spots bombarded with pCEV1/GFP-GUS (Figure 5c). It is worth pointing out that pCAMBIA1305.1 is a cloning vector for stable plant transformation. Just like other pCAMBIA vectors, the T-DNA region (containing selection marker nptII gene, expression cassette for GOI and reporter gene GUSPlus) in pCAMBIA1305.1 will integrate into the plant nuclear genome via a stable transformation process. However in the bombardment experiment followed by transient GUS staining assay, cytosolic GUS activity can also be detected when plant tissue is bombarded with pCAMBIA series (such as pCAMBIA1305.1), since the gus gene in the transformation vector is driven by non-tissue specific and constitutive CaMV 35S promoter. Particularly, CAMBIA's GUSPlus is a patented reporter gene isolated from Staphylococcus species with some properties that are superior to the conventional E. coli gusA gene (http://www.cambia.org). Previously, Ye et al. [31] reported that a chloroplast expression vector pHD203-GUS [carrying gus and chloramphenicol acetyltransferase (cat) genes driven by a double chloroplast psbA promoter fragment from pea (in opposite orientation)] and a positive nuclear control pBI505 (carrying a gus gene driven by a double CaMV 35S promoter plus a leader sequence from alfalfa mosaic virus) were separately bombarded into tobacco suspension cells, followed by GUS staining, and examined by microscopy. Cells bombarded with pBI505 showed high levels of GUS expression, as indicated by blue spots, and were distributed evenly throughout the cytosol of the transformed cells, whereas pHD203-GUS was expressed in chloroplasts. Chloroplast transient expression rates appeared to be 40-50-fold or even 300-400-fold lower than nuclear transformation rates [31]. Later, the same chloroplast expression vector pHD203-GUS together with positive (pBI221) and negative (pUC18) controls was used to develop an electroporation-mediated method for the study of foreign gene expression within spinach chloroplasts, and this study found that both GUS and CAT activities were detected in chloroplasts electroporated with pHD203-GUS but not with both controls; the expression of GUS protein in pHD203-GUS-treated chloroplasts was further confirmed by Western blot analysis [32]. Taken together, our observation in transient GUS expression is consistent with what was previously described under microscopy [31] and confirms that pCEV1/GFP-GUS is a functional vector.
Next, leaves from in vitro-grown plantlets of spinach (S. oleracea cv. Green Giant) were excised and bombarded with pCEV1/LCY, pCEV1/ZDS, and pCEV1/GFP-GUS individually. Almost all leaf segments became larger but turned brown in color after incubation for 1 month in selective medium, suggesting that the screening medium is effective ( Figure A1b). Eventually, a few green calli were identified (small panel in Figure A1b) from hundreds of explants after bombardment. Currently, plant regeneration from spectinomycin-resistant calli is in progress.

Construction of chloroplast expression vectors for tomato and N. benthamiana
In order to rapidly construct other chloroplast expression vectors for other plant species (i.e., tomato and N. benthamiana), a DNA fragment (1667 bp) containing expression cassettes of atpI promoter and cloning sites for GOI and rps16 terminator, followed by selection marker aadA gene, was amplified by a PCR strategy using a pCEV1 as a DNA template. After double digestion with SacI and PstI, the PCR product was ligated with pUC19 which was double digested with SacI and PstI, resulting in a new auxiliary vector pECaad (Figure 6a). Accuracy of the inserted DNA fragment has been confirmed by DNA sequencing. Thus, tomato LTR (containing partial 16S rRNA and trnI) can be inserted into the SacI site of the 5 0 terminus by using forward primer SacI+16S/trnI-F and reverse primer Le-16S/trnI +SacI-R and tomato DNA as a DNA template for PCR amplification; and tomato RTR (containing trnA and partial 23S rRNA) can be inserted into the PstI site of the 3 0 terminus by using forward primer Le-PstI+trnA/23S-F and reverse primer trnA/ 23S+PstI-R and tomato DNA as a DNA template for PCR amplification. As a result, chloroplast expression vector pCEV2 for tomato was constructed (Figure A2). To clone the GOI into pCEV2, the PCR fragment, carrying GFP:GUS fusion protein gene with NotI at one end and AscI at the other end, was inserted into pCEV2 which was double digested with NotI and AscI, resulting in pCEV2/GFP-GUS (Figure A2).
Complete plastid genome sequences have been reported for tobacco (N. tabacum; NC_001879), spinach (S. oleracea; NC_002202), and tomato (S. lycopersicum; NC_007898) but not for N. benthamiana. However, partial chloroplast sequence of N. benthamiana 16S-23S ribosomal RNA intergenic spacer including trnI and trnA genes has been determined (1874 bp; FJ217346). In order to prepare a longer targeting sequence (ca. 4 kb in total) in N. benthamiana, forward and reverse primers were designed within 16S rRNA and 23S rRNA, respectively, according to the plastid genome of N. tabacum, and total genomic DNA from N. benthamiana was used as a DNA template for PCR amplification. The PCR product was cloned and then sequenced. As expected, a longer targeting sequence (3893 bp) containing partial 16S rRNA, trnI, trnA, and partial 23S rRNA in N. benthamiana was obtained. As shown in Figure A3, 9 bp (GTTCGGCCT) insertion was conducted in N. benthamiana as compared to N. tabacum plastid genome. The same 9 bp insertion, located in the intron of trnI in N. benthamiana, has been reported [33]. In addition, two more nucleotides, indicated by the first and second arrow in Figure A3, were changed between N. tabacum and our N. benthamiana sequence. Moreover, five unknown nucleotides (labeled as "N") in the published N. benthamiana sequence (FJ217346) were determined in our N. benthamiana sequence. These five nucleotides in our N. benthamiana sequence were identical to N. tabacum (Figure A3).
Using the strategy in pCEV2, LTR and RTR from N. benthamiana were rapidly amplified by PCR and then inserted into pECaad, resulting in chloroplast expression vector pCEV4 for N. benthamiana (Figure A4). Similarly, GOI regarding gfp:gus, coat protein CP20 gene with or without 5 0 -and 3 0 -untranslated region (UTR) from BaMV satellite RNA was cloned separately into pCEV4 which was double digested with NotI and AscI, resulting in pCEV4/GFP-GUS, pCEV4/BSL6 (CP20 containing 5 0 -and 3 0 -UTR) and pCEV4/BaMV (CP20 only), respectively ( Figure A4). Therefore, the GOI can be inserted into our chloroplast expression vectors pCEV1, pCEV2, and pCEV4 specific for spinach, tomato, and N. benthamiana, respectively. Moreover, although the trnI/trnA fragment (here we used the longer sequence of partial 16S rRNA-trnI/trnA-partial 23S rRNA) is one of the most frequently used targeting region in plastid transformation [3,5], if a researcher wants to use another targeting region, such as rbcL/accD, rpl32/trnL, petA/psbJ, trnfM/trnG, trnN/trnR, atpB/rbcL, rps7/ndhB, or ycf3/trnS [5], such chloroplast targeting sequence can be amplified and cloned into our auxiliary vector pECaad (Figure 6a) which carries the expression cassette for the GOI and selection marker aadA on pUC19. We believe that our auxiliary vector pECaad will be of benefit for constructing different targeting regions for plastid transformation.
For selection in plastid transformation, aadA gene conferring spectinomycin resistance is widely used and is the most efficient selective marker [2]; however, nptII gene conferring kanamycin resistance has also been developed for tobacco plastid transformation [34]. Thus, besides the selection marker aadA gene which has been used in pCEV1, pCEV2, pCEV4, and their derivatives, we also constructed nptII gene as a selection marker for tomato plastid transformation. In brief, PCR-amplified BamHI-nptII-XbaI fragment was subcloned and replaced the BamHI-aadA-XbaI fragment in pECaad, generating another auxiliary vector pECnpt (Figure 6b) which carries an expression cassette for GOI and nptII selection marker on pUC19. Then, tomato LTR and RTR were sequentially cloned into pECnpt, generating a new chloroplast expression vector pCEV3 for tomato plastid transformation by using selection marker nptII gene (Figure A2). Similarly in pCEV2, the gene encoding GFP:GUS fusion protein was inserted into pCEV3 with double digestion by NotI and AscI, resulting in pCEV3/GFP-GUS (Figure A2).

Bombardment, selection, and verification of transplastomic N. benthamiana lines
To further evaluate the functional expression of our constructs, plasmids pCEV4/BaMV and pCEV4/BSL6 were bombarded separately into N. benthamiana. After bombardment, almost all leaf segments turned brown and finally died in NB selection medium containing 300 mg l À1 spectinomycin for 1-month culture. From a total of 344 leaf segments we used, only 1 shoot was regenerated using pCEV4/ BaMV as a transforming vector (Figure 7a), and also only 1 shoot was regenerated from a total of 340 leaf segments examined using pCEV4/BSL6 as a transforming vector (Figure 7b). These shoots were cut and transferred into MS basal medium without spectinomycin for rooting. Finally, two putative transplastomic plantlets, one from pCEV4/BaMV and another from pCEV4/BSL6, were obtained and transferred into a greenhouse. Thus, the transformation efficiency for both vectors pCEV4/BaMV and pCEV4/BSL6 was 0.3%.
For the putative transgenic line transformed by pCEV4/BSL6, genomic DNA was isolated from leaves of the putative line and wild-type N. benthamiana, and PCR amplification was carried out. A unique band of 837 bp, representing CP20 coding region flanked by 5 0 -UTR and 3 0 -UTR in the transforming vector pCEV4/BSL6, was detected in samples from the transplastomic line BSL6 as well as the positive control vector pCEV4/BSL6 (Figure 8b). No PCR product was detected in the WT plant (Figure 8b). To demonstrate the quality of DNA we isolated, a primer set amplifying a 342 bp fragment between trnA and 23S rRNA of plastid genome in N. benthamiana was designed (Figure 8a). As expected, a unique band (ca. 0.3 kb) was detected in our transplastomic plant BSL6, the WT and the positive control vector pCEV4/BSL6 (Figure 8c). To further confirm the integration site in the transplastomic line, forward primer TpsbA-F was designed within the terminator region of psbA in the vector pCEV4/BSL6, and reverse primer trnA-r was located at the trnA of the right targeting region in the vector pCEV4/BSL6. As predicted, a unique band of 351 bp was detected in transplastomic plant BSL6 but not in the WT (Figure 8d). In conclusion, the expression cassette carrying CP20 encoding gene (with 5 0 -and 3 0 -UTR) and aadA selection marker gene flanked by LTR (i.e., Nb 16S rRNA/trnI) and RTR (i.e., Nb trnA/23S) was indeed integrated into the trnI/trnA integration site by homologous recombination of plastid genome in transplastomic line BSL6. Unfortunately, serious infection by microorganisms occurred in this plant and it finally died.
For the putative transplastomic line which was transformed by pCEV4/BaMv, self-pollinated seeds from the T 0 transplastomic line BaMV as well as WT seeds were sterilized and then germinated on MS basal medium supplemented with 300 mg l À1 for 4 weeks. As shown in Figure 9b, all T 1 seedlings from the BaMV plant examined were resistant to spectinomycin, confirming an important characteristic of maternal inheritance in plastid transformation. By contrast, all seedlings from the WT were sensitive to spectinomycin (Figure 9a). To further confirm the transgene, three spectinomycin-resistant seedlings were randomly picked up from the selection plate, transferred onto MS basal medium without spectinomycin, and then grown in a greenhouse. Genomic DNA was isolated from these transplastomic T 1 as well as WT plants, and PCR analysis was then carried out (Figure 10).   A unique band of 729 bp, which is the size of CP20 encoding gene (without 5 0 -UTR and 3 0 -UTR) in the transforming vector pCEV4/BaMV, was detected in all T 1 transplastomic lines BaMV#1, BaMV#2, and BaMV#3 but not in the WT plant (Figure 10b). Again, the same PCR primer set was employed for the amplification between trnA and 23S rRNA in the plastid genome of N. benthamiana. As expected, a unique band of 0.3 kb was detected in all three transplastomic lines and the WT plant (Figure 10c). A unique band of 351 bp, representing the flanking sequence around the integration site at the RTR (i.e., Nb trnA/23S), was only detected in the three transplastomic lines but not in the WT sample (Figure 10d). Here we clearly demonstrated the expression cassette carrying CP20 encoding gene and aadA gene was integrated into the trnI/trnA site of the plastid genome in N. benthamiana via homologous recombination and the transgene was transmitted into progeny by maternal inheritance.

Bombardment in tomato
After bombardment, almost all tomato leaf explants turned brown and finally died in SL selection medium containing 300 mg l À1 spectinomycin (for pCEV2/GFP-GUS) or 50 mg l À1 kanamycin (for pCEV3/GFP-GUS). Eventually, at least two green calli were obtained for each construct ( Figure A5) from hundreds of explants examined. Currently, plant regeneration from spectinomycin-resistant calli is also in progress.

Conclusion
Here we constructed a series of chloroplast expression vectors specific for spinach (pCEV1), tomato (pCEV2 and pCEV3), and N. benthamiana (pCEV4). Common features of these vectors are described as follows: (1) sequence of cloning sites (NotI, SacII, AscI) for the GOI is located between plastid atpI promoter and plastid rps16 terminator; (2) selection marker aadA or nptII is located between plastid rrn promoter and plastid psbA terminator; (3) longer LTR (ca. 2 kb) contains partial 16S rRNA and trnI in the plastid genome; (4) longer RTR (ca. 2 kb) contains trnA and partial 23S rRNA in the plastid genome; (5) the integration site for the expression cassette is located in the intergenic spacer between trnI and trnA; and (6) the backbone plasmid is pUC19 (carrying ampicillin resistant marker gene).
Another feature of this study is the construction of the so-called "auxiliary vector" pECaad ( Figure 6a) and pECnpt (Figure 6b). Briefly, DNA fragment of expression cassette, which consists of (1) GOI cloning sites (NotI, SacII, AscI) flanked by atpI promoter and rps16 terminator and (2) selection marker gene aadA or nptII flanked by rrn promoter and psbA terminator, was ligated with pUC19 (carrying ampicillin resistant marker gene), generating pECaad or pECnpt. LTR can be inserted into the 5 0 terminus of the expression cassette, RTR can be inserted into the 3 0 terminus of the expression cassette, and GOI can be cloned into the cloning sites (NotI, SacII, AscI) within the expression cassette. Thus, a new vector suitable for other new plant species can be rapidly and efficiently constructed. We believe that our auxiliary vectors pECaad and pECnpt will aid the faster construction of new chloroplast expression vectors for other plant species. thank the Transgenic Core Facility of Academia Sinica for the biolistic bombardment assistance. We appreciate Miranda Loney for editing the manuscript. This work was financially supported by Academia Sinica of Taiwan to Dr. Kin-Ying To.   Sequence comparison for the targeting region (partial 16S rRNA, trnI, trnA, and partial 23S rRNA) between N. tabacum and N. benthamiana. N. tabacum sequence was extracted from NCBI Accession no. NC_001879. N. benthamiana sequence (3893 bp) was our PCR product. N. benthamiana* sequence (1874 bp) was the partial sequence of 16S-23S ribosomal RNA intergenic spacer from NCBI with the Accession no. FJ217346. A different nucleotide was indicated by red arrow and labeled with a red letter. Common region in these three sequences was highlighted in yellow color.