The codA gene as a negative selection marker in Citrus

The use of positive selectable marker genes is widespread in plant genetic transformation allowing transgenic cells to grow while repressing non-transgenic cells. Negative selectable markers, on the contrary, allow the repression or ablation of transgenic cells. The codA gene of Escherichia coli encodes cytosine deaminase that hydrolyzes 5-fluorocytosine (5-FC) into the cytotoxic compound 5 fluorouracil. We tested the transgenic expression of the bacterial codA gene in citrus as a conditional negative selection marker, with the goal of selecting against plant tissues in which a transgenic cassette has not been successfully removed. We developed transgenic citrus lines containing the selection cassette, codA::nptII, driven by double enhanced CaMV35S promoter, verified by Southern blot analysis, RT-PCR, DsRed expression and subjected these transgenic lines to a 5-FC sensitivity assay. We found that, while non-transgenic citrus were unaffected by the presence of 5-FC, all of the transformed lines displayed symptoms of toxicity, indicating that the codA gene could be used as a negative selectable marker in Citrus, for post-transformation detection of the removal of undesired sequences.


Background
The expression of a negative selectable marker gene causes either immediate or conditional cell death in transformed cells and therefore allows for the selection of cells lacking these marker genes (Koprek et al. 1999).
In several plant species, the codA gene from Escherichia coli, encoding cytosine deaminase, has been successfully used as a negative selection marker. The usefulness of codA as a conditional toxic gene has been explored in various Agrobacterium-mediated transformation protocols (Babwah and Waddell 2000;Dutt et al. 2008;Koprek et al. 1999;Rommens et al. 2004;Schaart et al. 2004). The gene product of codA deaminates cytosine to uracil, and also converts the uracil analog 5-fluorocytosine (5-FC) into 5-fluorouracil (5-FU). This substance is metabolized to 5-fluorouridine 5%-triphosphate and 5-fluoro-2%-deoxyuridine 5%-monophosphate, inhibiting both RNA and DNA synthesis, resulting in cell death (Andersen et al. 1989). Therefore, 5-FC is toxic for cells of several plant species transformed with the codA gene (Gallego et al. 1999;Kobayashi et al. 1995;Koprek et al. 1999;Perera et al. 1993). In this work, the codA coding sequence was isolated from the E. coli genome and cloned as a fusion to the nptII gene in the pCTAGV-KCN3 binary vector under control of the double enhance CaMV35S promoter. Agrobacterium-mediated transformation of citrus was carried out in order to evaluate the efficiency of selection using the codA gene as a negative marker. Nontransgenic and codA-transgenic internode segments were cultivated on selective medium containing 5-FC, showing that 5-FC did not affect the growth of non-transgenic citrus cells, whereas it inhibited shoot regeneration from codA-expressing transgenic cells. This positive-negative selectable marker may be applied to enhance recombinase-mediated cassette exchange for site-specific chromosomal integration in citrus (Wang et al. 2011).

Plant material
In vitro-grown etiolated epicotyls segments of 'Carrizo' citrange (Citrus sinensis × Poncirus trifoliata) and Open Access *Correspondence: James.Thomson@ars.usda.gov 2 USDA-ARS Crop Improvement and Utilization Research Unit, Albany, CA 94710, USA Full list of author information is available at the end of the article 'Hamlin' sweet orange [Citrus sinensis (L.) Osbeck.] were used as source of explants. Surface-sterilized seeds were inoculated in solution containing 12 ml of hormone-free half-strength MS basal medium (Murashige and Skoog 1962), supplemented with 3% sucrose and solidified with 8.0 g l −1 agar (Sigma-Aldrich) for germination. The test tubes were capped with polypropylene closures. The cultures were incubated in a growth chamber at an average temperature of at 27 ± 1°C under dark conditions for 6 weeks. The epicotyl portions of etiolated seedlings were cut transversally into 0.8-1 cm segments and used for transformation experiments.

Bacterial strain and vector
Plasmid pCTAGDV-KCN3 (Figure 1a) was transformed into A. tumefaciens EHA 105 (Hood et al. 1986) using freeze-thaw method (Höfgen and Willmitzer 1988). The plasmid pCTAGDV-KCN3, based on the pCam-bia390 vector backbone, contains the double enhanced 35S promoter constitutively expressing fusion gene codA::nptII a bifunctional selectable marker, and a DSRed visual reporter gene constitutively expressed by the GmUbi3 promoter (Hernandez-Garcia et al. 2010). The pCTAGDV-KCN3 binary vector is a new plant genetic transformation vector contains a recombinase recognition site platform that enables recombinase mediated cassette exchange (RMCE) to target gene(s) of interest and remove unneeded DNA sequences, such as antibiotic resistance genes (Wang et al. 2011).

Plant transformation and regeneration of transgenic plants
Epicotyl tissue of Carrizo citrange was inoculated for 15 min with an overnight culture of bacteria diluted to OD 600 = 0.4 and blotted on sterilized filter paper to remove excess bacteria in suspension. The infected epicotyl tissue was co-cultivated for 3 days on MS basal medium containing 3% sucrose, 1 mg l −1 BAP, 100 mg l −1 acetosyringone, 8.0 g l −1 agar and incubated at an average temperature of 24 ± 1°C in the dark. After 3 days of cocultivation, the explants were transferred to MS medium containing 300 mg l −1 timentin (ticarcillin and clavulanate), 250 mg l −1 cefotaxime, and 100 mg l −1 kanamycin. After 45 days in culture on kanamycin selection media, putatively transformed shoots (3-5 cm in height) were transferred to root-induction medium (RIM), consisting of ¼ MS strength medium, 2% sucrose, 0.25% Gelrite, 2.5 mg l −1 IBA, 0.5 mg l −1 NAA, and 0.0025 mg l −1 spermidine. After rooting, plants with developed root systems were transferred to sterile soil cones, covered with plastic bags and gradually exposed to ambient humidity in a growth chamber, over a period of 15 days. Acclimated plants were transferred to greenhouse for maturation.

Molecular analyses
The genomic DNA was isolated from leaves of transgenic and non-transgenic plants, as previously described (Štorchová et al. 2000). The concentration of DNA was measured using a Nanodrop 2000c (Thermo Scientific). For each sample, 100 ng of genomic DNA in 25 µl volume was used in a PCR with the primers codAORF70F60 and codAORF1137R60 primers (5′-CAAGACCCTTCCTCT ATATAAG-3′ and 5′-CGAGTTCATAGAGATAACC TTC-3′, respectively) to detect codA expected fragment size 1,067 bp. As a control, DNA from non-transformed cv. Carrizo was amplified.

Southern blot
For Southern-blot analysis, 5 µg of genomic DNA of non-transgenic and transgenic plants were digested with EcoRI for 6 h at 37°C and separated by electrophoresis on a 0.8% (w/v) agarose gel. The DNA was then transferred to a Hybond-N membrane (Amersham) and hybridized with the 32 P-labelled codA sequence (produced by Promega TaqTM polymerase using primers codAORF70F60 and codAORF1137R60).

Expression of DSRed
Expression of Red fluorescent marker genes in the transgenic and non-transgenic shoots were examined via Olympus Provis AX70 fluorescent microscope (Olympus) at 10× magnification and the appropriate filter for detection of the red fluorescence of the DSRed gene, which has an excitation maximum at 545 nm and emission maximum at 600 nm. The pictures were taken using a Q Color 5 Olympus camera with Q Capture software.

5-FC sensitivity assay
Transgenic lines confirmed by PCR, Southern blot, and RT-PCR were propagated and 18-month-old internodes were then cultivated in SIM supplemented with Kanamycin (Kan) and different concentration of 5-FC, in order to study both positive and negative selection phenotypes. Internode segments of 0.8-1 cm length from transgenic and non-transgenic plants were surface sterilized (70% ethyl alcohol for 120 s followed by 25% bleach for 15 min and washed three times with water), placed in Petri dishes 120 × 20 mm containing 25 ml of shooting induction media (SIM) based on the nutrients and vitamins of WPM medium (Lloyd and McCown 1980) supplemented with 2 mg l −1 BAP with five different concentrations of 5-FC (Sigma) (0, 50, 75, 100, 200, or 400 mg l −1 ), kept in the dark for 3 weeks at 26°C after which they were transferred to 16/8 h light/dark cycles at 27 ± 1°C. Nontransgenic plants were also tested for their response to various 5-FU (Sigma) concentrations (10, 20, 40, 80 and 100 mg l −1 ). 5-FC and 5-FU were dissolved in Milli-Q water at 65°C at concentration of 10 g l −1 . These solutions were filter sterilized and immediately added to SIM medium at ~65°C to indicated dilutions. Counts of percentage of regeneration and numbers of shoots per explants were taken after 70 days. Results were expressed as the mean and compared using the Tukey's multiple comparison test at 5% probability for 5-FC variables studied.

Rooting assay using selective and non-selective media with 5-FC
Adventitious shoots of non-transgenic and transgenic line codA2 were grown in SIM, and then transferred to root-induction medium (RIM), as previously described, containing 0, 50, 75, 100, 200 or 400 mg l −1 FC. Frequency of rooting was scored after 45 days in RIM.

Selection and regeneration of transgenic plants
We have designed a transformation construct based on the pCambia390 backbone binary vector. The T-DNA contains a neomycin phosphotransferase II gene (nptII), which confers kanamycin resistance in plants as a positive selectable marker and the cytosine deaminase gene (codA) as the conditional negative selectable marker. The codA and nptII coding sequences are translationally fused as a single cassette and expressed by the ds35S promoter and terminator. The DSRed gene was included as a viable visible marker for early transformation detection (Figure 1a). A total of 335 transgenic citrus lines from Carrizo and Hamlin were generated after 8-10 weeks of culturing epicotyl explants on regeneration/selection medium with kanamycin (100 mg l −1 ). These shoots grew normally and rooted.

Molecular characterization of the regenerated plants
Putative transgenic plant lines were analyzed using the codA specific primer set to determine the presence of the negative marker (not shown). T-DNA copy number was determined using Southern blot analysis of an EcoRIdigested genome from 123 transgenic lines in Carrizo and 15 in Hamlin using a P 32 -labelled codA sequence as a probe. No signal was detected in the DNA from nontransgenic plants included as a negative control. Since there is only one EcoRI site in the transformed vector a second EcoRI site must be derived from the citrus genomic DNA and diagnostic for the number of T-DNA cassettes present. The number of T-DNA cassettes varied from one to five with most of the lines tested having one or two copies (examples in Figure 1b). Sixty-five plants from Carrizo and nine from Hamlin showed a single copy T-DNA integration site. Transgenic plants with single insertions were selected for further research.
Four lines with a single copy T-DNA insert (codA lines# 1, 2, 4, and 7) and four with multiple copy inserts (codA lines# 3, 5, 6, and 8) were analyzed by Reverse-Transcriptase-PCR (RT-PCR). The results of RT-PCR using codA specific primers showed that the transgene is transcribed in the transgenic plants (Figure 1c). Both transgenic and non-transgenic plants were examined by fluorescent microscope. Red fluorescence resulting from production of the DSRed protein was clearly visible in the transgenic lines. While the red fluorescent signal strength varied between lines, all of the transformed lines exhibited activity, while fluorescence was absent in nontransgenic citrus (Figure 1d).

Negative selection assay
To determine whether the 5-FC CodA metabolic product, 5-FU, was an effective toxic agent on citrus internode segments, non-transgenic tissue was placed on media containing different concentrations of 5-FU (Table 1). A concentration of 20 mg l −1 resulted in strong reduction of the percentage of regeneration as well as number of shoots per explants in non-transgenic internode tissue. The resulting shoots showed abnormal leaf development and severe reduction in shoot length. At concentrations of 80 and 100 mg l −1 there was no detectable regeneration.
The effects of negative selection were examined on three transgenic Carrizo lines with one, two or three copies of the transgene cassette (codA lines# 2, 3, 6, respectively; Figure 1c), as well as on Carrizo non-transgenic. Eighteen month old internode tissue from the transgenic and non-transgenic lines were subjected to five levels of 5-FC. At 50 and 75 mg l −1 of 5-FC significantly affected shoot growth of codA expressing internode tissue in all three transgenic lines tested, causing the percentage of regeneration and number of shoots per explants to decrease by about 50 and 75%, respectively (Table 2). Non-transgenic internode tissue regenerated and produced a normal number of green and healthy shoots per explant at these concentrations.
At 100 mg l −1 of 5-FC, non-transgenic internode tissue was again similar to untreated plant material, while the transgenic internode tissue completely failed to regeneration shoots. However, growth and development of nontransgenic tissue was inhibited at 400 mg l −1 5-FC, giving an upper limit to the use of this compound as a selective agent with citrus. Figure 2 summarizes results observed

Table 1 Negative selection determination with 5-fluorouracil
Regeneration frequency and number of shoots per internode explant of non-transgenic (NT) citrus in regeneration medium supplemented with various concentrations of the toxic metabolite 5-fluorouracil (5-FU). Regeneration frequency and number of shoots per explants were taken after 70 days.
Data are the means of three independent experiments. Values are regeneration % (1) and number of shoots per explants (2) ±standard deviation between repetitions.  using the different levels of 5-FC on shoot growth of codA expression transgenic internode tissue compared to non-transgenic. In order to test rooting induction in the presence of 5-FC, shoots from both non-transgenic and codA transgenic line #2 (codA2) were placed on rooting induction media (RIM) containing various concentrations of 5-FC, as indicated in Figure 3a. At 400 mg l −1 of 5-FC in RIM, few shoots from the transgenic line produced roots (4.4%), and when roots were produced they were small and stunted, and leaves appeared yellow (chlorotic) when compared to non-transgenic under similar conditions (Figure 3b). Use of 200 mg l −1 5-FC permitted rooting of 25% of the codA2 transgenic shoots. The 5-FC concentrations 100, 75, and 50 mg l −1 showed little or no marked difference in rooting inhibition between the transgenic and wild type tissues. Thus, we demonstrate that the codA gene can be used as an effective conditional negative selection marker in citrus cultivar Carrizo at a concentration of 5-FC of 100 mg l −1 for shoot regeneration while a concentration of 400 mg l −1 was necessary to inhibit rooting.

Discussion
We have shown that codA is an effective negative selection marker. We produced and molecularly characterized transgenic citrus expressing one or more copies of the codA gene without any observable phenotypic abnormalities.
Furthermore, we have demonstrated that non-transgenic citrus internode tissue is sensitive to the presence of 5-FU (Table 1; Figure 2) and insensitive to 5-FC at

Table 2 Negative selection determination with 5-fluorocytosine
Internodes from non-transgenic and three transgenic lines, carrying the codA::nptII gene, were placed in regeneration medium supplemented with various concentrations of 5-FC. Regeneration frequency and number of shoots per explants were taken after 70 days.
Data are the means of three independent experiments. Values are regeneration % (1) and number of shoots per explants (2) ±standard deviation between repetitions. Values in a column followed by a common letter are not significantly different according to Tukey's multiple range test (P < 0.05). Lines codA2, codA3, and codA6 have one, three, and two transgene cassette copies, respectively.  Internodes segments were placed on a Petri dish with shoot-induction medium containing various 5-FC concentrations. The explants were grown for 21 days under dark condition at 26°C and then transferred to light/dark condition at 27 ± 1°C.

5-FC selection
concentrations below 200 mg l −1 . In contrast, transgenic plants expressing the codA gene showed high sensitivity to 5-FC (Table 2). These results are consistent with previous publications using 5-FC as a selective agent in transgenic L. japonicas leaves (Stougaard 1993), N. tabacum leaves (Schlaman and Hooykaas 1997) suspension cells (Gallego et al. 1999), H. vulgare scutellar tissue (Koprek et al. 1999), and B. napus cotyledons (Babwah and Waddell 2000;Kobayashi et al. 1995;Stougaard 1993). We observed that 100 mg l −1 of 5-FC was sufficient to inhibit shoot regeneration from codA expressing transgenic tissue whereas non transgenic tissue was unaffected for regeneration at concentrations below 200 mg l −1 . However, use of 5-FC for rooting inhibition required 400 mg l −1 for effective selection of codA transgenics.
In conclusion, this work demonstrates that the codA gene can be successfully used as a negative selectable marker in transgenic citrus, and is now available for use as a tool to facilitate precise genome modification strategies.