In vitro plant regeneration and Agrobacterium-mediated genetic transformation of a carnivorous plant, Nepenthes mirabilis

In nutrient-poor habitats, carnivorous plants have developed novel feeding strategies based on the capture and digestion of prey and the assimilation of prey-derived nutrients by specialized traps. The Nepenthes genus, comprising nearly 160 species, presents a remarkable pitcher-shaped trap, leading to great interest among biologists, but the species of this genus are listed as threatened. In this work, we developed a protocol for reproducing Nepenthes mirabilis through shoot regeneration from calli. The cultivation of stem segments of N. mirabilis on MS medium containing thidiazuron induced organogenic calli after 10 weeks. Subcultured calli exposed to 6-benzylaminopurine showed shoot regeneration in 3 weeks with considerable yields (143 shoots/g of calli). Excised shoots transferred to medium with indole-3-butyric acid allowed rooting in 4 weeks, and rooted plantlets had a 100% survival rate. Based on this method, we also developed an Agrobacterium-mediated genetic transformation protocol using calli as explants and ipt as a positive method of selection. Twelve weeks post infection, regenerated shoots were observed at the surface of calli. Their transgenic status was confirmed by PCR and RT-PCR. In conclusion, this study provides an efficient method for regenerating Nepenthes and the first protocol for its stable genetic transformation, a new tool for studying carnivory.

Shoot regeneration. Some organogenic calli (200-400 mg) were subcultured on medium supplemented with BAP (0.1-2 mg l −1 ), kinetin (0.1-2 mg l −1 ) and TDZ (0.1-2 mg l −1 ) ( Table 2). Three weeks after transfer to the regeneration medium, we observed that the surface of calli exposed to light became green and showed a granular appearance corresponding to shoot apical meristems (Fig. 1C), which later generated foliar primordia (Fig. 1D,E) in the presence of BAP and kinetin. The regeneration efficiency and the number of shoots per callus varied depending on the type and concentration of cytokinin used. Media devoid of growth regulators or supplemented with BAP (0.1-2 mg l −1 ) or kinetin (0.1-2 mg l −1 ) produced the highest frequency of shoot regeneration (87-100%, Table 2, Fig. 1F-G). Moreover, exposure to BAP (0.5-2 mg l −1 ) led to the regeneration of a considerable number of shoots (115-143 shoots/gr of calli, Table 2).
Rooting. Shoots of 0.5-1 cm generated on 1 mg l −1 BAP medium were transferred to medium supplemented with various concentrations of IAA (indole-3-acetic acid), IBA (indole-3-butyric acid) and NAA (0.1-2 mg l −1 ) ( Table 3). Root primordia were observed 4 weeks after transfer to the rooting medium. Regardless of the concentration that was used (0.1-2 mg l −1 ), IAA and IBA were clearly more efficient in promoting root induction, with success rates between 92 and 100% (Table 3). IBA at 0.5-2 mg l −1 allowed the development of a rooting system presenting a high number of roots (± 11 roots/shoot) with a size varying between 1.3 and 1.7 cm, divided into many ramifications (4-6 ramifications/root, Table 3, Fig. 1H). The plants grown on this medium also produced large aerial organs of between 5 and 7 cm. In contrast, plantlets rooted on the NAA medium grew more slowly Table 1. Effects of PGRs on organogenic callus induction from nodal stem segments after 10 weeks (% of callus induction). Ten explants were cultured in separated plates for each treatment. Experiments were repeated 3 times. Means followed by the same letters are not significantly different (P < 0.05). www.nature.com/scientificreports/ because of a limited rooting system. Regenerated plantlets with well-developed roots were acclimated in a greenhouse with 100% efficiency for plants rooted in medium containing IBA (Fig. 1I).
Genetic transformation and ipt-mediated positive selection. The establishment of an Agrobacterium-mediated transformation protocol requires several parameters to be set, such as the conditions for the selection of transformed cells. Preliminary experiments performed on N. mirabilis calli provided evidence that the necrosis of untransformed cells on selective medium (containing classical herbicides or antibiotics) led to the death of neighboring cells. We therefore switched to a positive selection system based on the production of isopentenyl transferase (ipt). This enzyme is reported to increase the production of endogenous cytokinins and lead to shoot formation. Undifferentiated callus fragments (200-400 mg) were used as explants for genetic transformation. Twentythree series of transformation experiments including 20 calli each were conducted as described in the Materials and Methods section. The transformations were performed with Agrobacterium containing pGWB2-ipt or pGWB2-mgfp5 as a negative control. To prevent shoot formation and maintain the cells in their undifferentiated state, calli dipped in an Agrobacterium suspension were cultured on medium supplemented with TDZ. In addition, low concentrations of hygromycin (1.5 mg l −1 ) were used to slow down the cell growth of nontransformed cells. Under these conditions, we assumed that the plantlets that appeared at the surface of the calli were able to www.nature.com/scientificreports/ express the ipt gene. Ten to twelve weeks postinfection, we observed spontaneously regenerated shoots at the surface of calli transformed with pGWB2-ipt (Fig. 2B), whereas no shoots were observed on any calli infected with A. tumefaciens harboring the pGWB2-mgfp5 gene ( Fig. 2A).
To investigate the integration of the ipt gene in the genome, we collected a single shoot per callus and carried out specific PCR amplification of the transgene for each of them. The expected 720 bp-long amplicon was observed in 60.25% of the plants tested (Fig. 3B). The variation of the band intensity in comparison to the Nepenthesin 2 gene (Fig. 3A) could, however, reflect a potential chimeric state of the plants.
The selected ipt-transformed plants further developed a typical cytokinin-overproducing phenotype ( Fig. 2D-H) in comparison to wild-type plants (Fig. 2C). For several transformation events, the intensive reproduction of plantlets was observed at the crown level of shoots separated from calli. This led to the development of a pseudocallus that was entirely recovered by a large number of small plants displaying restrained stem elongation Table 2. Effects of different concentrations of cytokinins on the regeneration of shoots from organogenic calli produced from nodal stem segments. Ten callus fragments produced under the same growth regulator combination were cultured in separate plates for each treatment. Experiments were repeated three times. Means within a column followed by the same letters are not significantly different at the 5% level.  Table 3. Effects of different concentrations of auxins on rooting efficiency and the morphology of the root system. Ten shoots regenerated under the same is the second key growth regulator combination were cultured in separate jars for each treatment. Experiments were repeated three times. Means within a column followed by the same letters are not significantly different at the 5% level. www.nature.com/scientificreports/ ( Fig. 2D-F). For other plants, we observed reduced apical dominance and severe axillary branching at the apical and nodal levels, reflecting a typical phenotype observed in the presence of high concentrations of cytokinins ( Fig. 2G,H). The rate of root initiation was extremely low for most transformation events despite exposure to 1 mg l −1 IBA for 18 months. This observation is consistent with the overproduction of endogenous cytokinins. Only 25 of 91 transgenic shoots developed adventitious roots, grew to the plantlet stage and were successfully acclimated. Reverse transcription-PCR experiments were performed on the transgenic plants to highlight the correlation between the expression of the ipt gene and the cytokinin-overproducing phenotype. We highlight the presence of a 430 bp-long amplicon corresponding to the expected ipt fragment of 6-to 8-month-old plants (Fig. 4). The intensity of the amplified DNA fragment could not be related to a specific phenotype. The differences between the 12 plants that we tested could therefore be related to the insertion site of the gene in the genome or the possible chimeric state of the transgenic plants.

Discussion
In this work, we developed a method for regenerating and genetically transforming Nepenthes mirabilis under in vitro conditions. In the first part of the study, we identified a cocktail of growth regulators based on TDZ, BAP and IBA necessary for successful plant regeneration through indirect organogenesis allowing the regeneration of up to + /− 140 shoots/g of calli.
Callus cultures could be established from multinodal stem tissues exposed to different concentrations of TDZ with a 30-84% rate of success. This plant growth regulator (PGR) allows calli to be maintained in their undifferentiated state by inhibiting shoot regeneration. TDZ seems to have the ability to switch the internal hormonal balance from nodal meristem production toward a simple cell proliferation program. More than 800 articles report the use of TDZ as a cytokinin-like PGR that has been widely used for the induction of plant regeneration 25 . The potency of TDZ has been demonstrated for the in vitro propagation of many recalcitrant, www.nature.com/scientificreports/ woody and legume species. For example, it has been used for the culture of Echinacea purpurea L. 25 , Vitex trifolia L. 26 , Curcuma longa L. 27 , Bauhinia tomentosa L. 28 and Psoralea corylifolia 29 . In contrast to other cytokines, TDZ is resistant to endogenous cytokinin oxidases and is therefore rather stable in tissue culture. It has also been shown that it can inhibit the activity of these enzymes, resulting in the accumulation of endogenous purine cytokinins (reviewed by Dewir et al. 30 ). Finally, TDZ was reported to maintain stable callus culture since it does not alter the growth rate and shoot formation is repressed at least until 10 weeks after induction. Dewir et al. indicated that plant tissues exposed to TDZ for a long duration are subjected to an overdose, resulting in the inhibition of shoot proliferation 30 . BAP is the second key growth regulator that plays an important role in the method we developed. This molecule is reported to be used for the activation of seed germination with an efficiency of 26-97% 31,32 . In addition to this property, this growth regulator causes an increase in the development of shoot apical meristems on callus surfaces when added to the culture medium (0.5-2 mg l −1 ). This observation is consistent with many reports concerning the in vitro culture of other Nepenthes species. Indeed, BAP seems to be one of the most efficient cytokinins for the generation of shoots from nodal stem fragments or the apical region. The success of this induction method is reported to be between 90 and 95% [33][34][35][36] . We estimate that the ability of BAP to activate meristems  www.nature.com/scientificreports/ and regenerate several shoots from the same node leads to a 2.5-fold increase in shoot apical meristems on the callus surface in comparison to culture conditions without growth regulator. Finally, the third growth regulator necessary in our culture method is IBA. When shoots were cultured on medium containing 0.5-2 mg l −1 IBA, we observed nearly 100% rooting. This PGR allows the generation of multiple large primary roots displaying substantial ramification.
Today, 122 Nepenthes species are included in the International Union for Conservation of Nature (IUCN) red list of threatened species. Among these species, 14 are classified as endangered, and 10 exhibit a critical status. One way to save these plants is to reproduce them in vitro. Several relevant reports are already available. In vitro reproduction for the large-scale propagation of N. khasiana, N. mirabilis and N. macfarlanei has been performed by seed germination and tissue culture [31][32][33][34][35] . Germination under in vitro conditions is difficult and depends on the freshness of seeds and their contamination state. Therefore, micropropagation from nodes is mostly described in the literature as an interesting option for rapid mass reproduction and large-scale propagation. The method we describe in this report seems to be an interesting approach for efficiently reproducing N. mirabilis. It could be considered for application to other endangered Nepenthes species for their conservation, although it might be necessary to determine the appropriate PRG concentrations for each of them. Although the approach is very efficient and fast, side effects such as somaclonal variations should potentially be considered. Such modifications were observed by Devi and collaborators for N. khasiana. These authors highlighted cytological variations in three consecutive regenerations of N. khasiana from nodal stem fragments 37,38 . Since TDZ can cause abnormalities in plant tissue cultures, such an evaluation should be performed under our method.
In the second part of our study, we set up an Agrobacterium-mediated genetic transformation protocol for Nepenthes mirabilis. Transgenic plants were obtained from calli using a positive selection strategy based on the expression of isopentenyl transferase. We observed that when we used this approach, nearly 60% of the regenerated shoots were transgenic. Such a strategy might therefore be considered for gene functional analysis purposes, and any gene of interest might be included in the T-DNA of a binary vector under the control of the P35S promoter containing the ipt gene. This efficient method applied in N. mirabilis was previously used to generate luciferase-or gus-transgenic tobacco plants 39,40 . Although this method is efficient, it could still be improved. For example, our PCR analyses seem to highlight chimerism events in the transgenic plants regenerated from callus culture despite the inclusion of hygromycin in the medium. Ebinuma and colleagues demonstrated that ipt under the control of the P35S promoter leads to cytokinin overproduction in transformed cells. The hormone likely diffuses to wild-type neighboring cells and induces the generation of chimeric plants or wild-type shoots near transgenic shoots 41 . Kunkel et al. (1999) proposed that the number of nontransgenic regenerants could be reduced by exposing tissue to high ratios of auxins favorable to root regeneration and suppressing shoot regeneration to counterbalance cytokinin effects 39 . Such a strategy could be applied to N. mirabilis calli after the transformation step.
Another drawback of this strategy is the generation of typical detrimental phenotypes related to cytokinin overproduction, characterized by strongly restrained root initiation and shoot elongation. Therefore, most of the ipt-transformed shoots could not produce roots in the rooting medium and could not be acclimatized. However, several solutions have been described in the literature to limit this negative effect. The first consists of fusing the ipt gene with an inducible promoter, such as a copper-42 , ethanol-43 , tetracycline- 44,45 or dexamethasone (Dex)-inducible promoter 39 . Another possibility might be to remove the ipt gene using the MAT (multi-autotransformation) vector system. This system allows the excision of the ipt gene though site-specific recombination induced by a recombinase of the R/RS system 41,46,47 . This approach has been successfully used in several plant species, such as tobacco 41,41 , hybrid aspens 48 , rice 49 , Nierembergia 40 , apricot 50 , white poplar 51 , citrus 52 , cassava 53 , Kalanchoe blossfeldiana 54 , petunia 55 , potato 56 and tomato 57 . An alternative method relying on the Cre/loxP siterecombination system also allows the generation of marker-free transgenic plants, as described for citrus 58 .
The study described in this article provides some new perspectives related to the preservation of Nepenthes species. The efficient in vitro method that we developed provides an interesting tool for reproducing these plants and subsequently contributing to their restoration in the context of a general decrease in biodiversity. The work also provides a new tool for investigating the molecular mechanisms involved in plant carnivory, such as those related to the formation of the fascinating traps of these plants. For example, overexpression or genomic editing methods might help to provide physiological validation of genes involved in pitcher development, such as the recently described ASYMMETRIC LEAVES1 and REVOLUTA genes 59 . Genetic and proteomic resources have greatly increased over the last 10 years, including a multitude of RNA-seq and genome libraries from different Nepenthes species. All of these resources available in public databases can now be fully utilized for understanding how these plants originated and evolved.

Materials and methods
Plant material and tissue culture. In vitro N. mirabilis cultures were established from fresh seeds supplied by Districarnivores (www.distr icarn ivore s.com). Seeds were sterilized by total immersion in a diluted commercial bleach solution containing 0.25% sodium hypochlorite for 5 min and washed three times with sterile water. After a drying step on sterile paper, the seeds were sown in Petri dishes of 58
The A. tumefaciens GV310I strain transformed with the recombinant pGWB2-ipt or pGWB2-m-gfp5-ER plasmid was cultured in liquid YEB medium containing 30 mg/l rifampicin, 30 mg/l gentamycin and 30 mg/l kanamycin at 28 °C for 2 days at 120 rpm. Four hours before plant transformation, 100 µM acetosyringone was added to the bacterial cultures from a − 20 °C-stored stock solution of 1 M in concentration. The bacteria were pelleted by centrifugation for 12 min at 3500×g and resuspended at a cell density of OD 600 0.7 ± 0.1 in liquid BM.
Genetic transformation of N. mirabilis. Undifferentiated callus fragments (200-400 mg) were dipped in an A. tumefaciens culture for 15 min, briefly dried on sterile filter paper and transferred to solidified BM supplemented with 1 mg l −1 TDZ. After 3 days of cocultivation (16 h/8 h day/night photoperiod at a temperature of 23 °C), the calli were transferred to BM supplemented with 1 mg l −1 TDZ, 1.5 mg l −1 hygromycin and 200 mg l −1 cefotaxime. To favor rooting, the regenerated plantlets were separated from the calli and transferred to BM supplemented with 1 mg l −1 IBA, 1.5 mg l −1 hygromycin and 200 mg l −1 cefotaxime.

Statistical analysis.
For the establishment of regeneration tests, the experiments were performed under a randomized design. Ten explants were cultured in separated plates for each treatment. Experiments were repeated 3 times. The data collected were subjected to Student's t-tests (P ≤ 0.05).  Figure 5. Schematic representation of the T-DNA used in this study, pGWB2-ipt (A) and pGWB2-gfp (B). RB: right border; LB: left border; Tnos: terminator of nopaline synthase gene; Pnos: promotor of nopaline synthase gene; P35S: CaMV 35S promoter; nptII: neomycin phosphotransferase gene; hpt: hygromycin phosphotransferase; mgfp5: green fluorescent protein gene; ipt: isopentenyl transferase gene.