Constitutive expression of transgenes encoding derivatives of the synthetic antimicrobial peptide BP100: impact on rice host plant fitness

Background The Biopeptide BP100 is a synthetic and strongly cationic α-helical undecapeptide with high, specific antibacterial activity against economically important plant-pathogenic bacteria, and very low toxicity. It was selected from a library of synthetic peptides, along with other peptides with activities against relevant bacterial and fungal species. Expression of the BP100 series of peptides in plants is of major interest to establish disease-resistant plants and facilitate molecular farming. Specific challenges were the small length, peptide degradation by plant proteases and toxicity to the host plant. Here we approached the expression of the BP100 peptide series in plants using BP100 as a proof-of-concept. Results Our design considered up to three tandemly arranged BP100 units and peptide accumulation in the endoplasmic reticulum (ER), analyzing five BP100 derivatives. The ER retention sequence did not reduce the antimicrobial activity of chemically synthesized BP100 derivatives, making this strategy possible. Transformation with sequences encoding BP100 derivatives (bp100der) was over ten-fold less efficient than that of the hygromycin phosphotransferase (hptII) transgene. The BP100 direct tandems did not show higher antimicrobial activity than BP100, and genetically modified (GM) plants constitutively expressing them were not viable. In contrast, inverted repeats of BP100, whether or not elongated with a portion of a natural antimicrobial peptide (AMP), had higher antimicrobial activity, and fertile GM rice lines constitutively expressing bp100der were produced. These GM lines had increased resistance to the pathogens Dickeya chrysanthemi and Fusarium verticillioides, and tolerance to oxidative stress, with agronomic performance comparable to untransformed lines. Conclusions Constitutive expression of transgenes encoding short cationic α-helical synthetic peptides can have a strong negative impact on rice fitness. However, GM plants expressing, for example, BP100 based on inverted repeats, have adequate agronomic performance and resistant phenotypes as a result of a complex equilibrium between bp100der toxicity to plant cells, antimicrobial activity and transgene-derived plant stress response. It is likely that these results can be extended to other peptides with similar characteristics.


Background
Antimicrobial peptides (AMPs) are short sequence peptides, normally less than 50 amino acid residues, reported in living systems. They are components of the defense system against pathogens in plants and animals or are produced by microorganisms in antibiosis processes (see reviews in [1][2][3] bacteria; [4,5] fungi; [6,7] insects; [8][9][10] amphibian and mammals, and [11] plants). Around 1,000 AMPs have been reported [12]. They can structurally be linear peptides (often adopting α-helical structures); cysteine-rich open-ended peptides with disulfide bridges; cyclopeptides forming a peptide ring, or pseudopeptides. AMPs offer major perspectives as a novel class of therapeutic agents, especially against fungal infections and antibiotic-resistant bacterial pathogens in humans and animals [7,9]. This great potential extends to plant disease-protection products [13][14][15], as substitutes of antibiotics in animal feed, biopreservatives in food, cosmetics and biomaterials, and as antifouling agents [16,17]. AMPs have proved successful as biopesticides, with commercial development of several microorganisms secreting these compounds [14].
In recent years, novel peptides have been designed, based on natural AMPs, with the aim of optimizing the activity against selected target pathogens (including microorganisms against which no AMP or antibiotic are known) while decreasing toxicity to non-target organisms and increasing stability. Short truncated compounds (minimal domain), chimerical constructions and improved sequence analogs have been reported. Examples are mellitin derivatives blocking plant viruses [18], the anti-fungal and anti-bacterial lactoferricin B derivatives [13], antifungal cecropin A and cecropin A-mellitin derived peptides [19][20][21], and the de novo designed antifungal hexapeptide PAF26 [22,23] and bactericide cyclic decapeptide BPC194 series [19,20].
Combinatorial chemistry approaches have been used to assist the design of new AMPs with superior properties.
The CECMEL11 peptide library, a 125-member linear undecapeptide library contains groups of sequences with high activity against a number of reporter bacterial and fungal phytopathogenic species, several also exhibiting lowsensitivity to protease degradation and hemolytic activity [47,48]. These peptides were cecropin A-mellitin hybrids and had the structure of an amphipathic α-helix with strong positive charge at neutral pH that can facilitate electrostatic attraction to phospholipid membranes of the target microorganisms, prior to insertion of the hydrophobic face into the membrane bilayer [49,50]. In particular, BP100 (KKLFKKILKYL-NH2) displays strong, selective bactericidal activity against three plant pathogenic bacteria at micromolar concentrations, with poor antifungal properties that inhibit infections by Xanthomonas vesicatoria in pepper, Erwinia amylovora in apple and Pseudomonas syringae in pear [47]. The efficacy of control is comparable to standard antibiotics and it is highly biocompatible, as assessed by acute oral toxicity tests in mice [51].
The envisaged phytosanitary applications of the BP100 series of peptides makes its expression in plant systems relevant, either as potent tools to confer plant phenotypes resistance to bacterial and/or fungal pathogen species, or as biofactories of plant protection products. Phytosanitary use of these peptides against bacterial diseases of plants is limited by the high cost of production. Large-scale chemical synthesis of peptides above around 6 amino acids is only economically viable for applications of very high added value; in view of its low toxicity against animal models, the use of GM plants as AMP molecular farms could putatively be an economic alternative. Production of a number of proteins using plants as biofactories has been reported in pharmaceutical applications [52][53][54], and a number of companies are currently using different approaches with various proteins [55].
However, expression of AMPs in plants requires specific strategies, due to their particular properties, and has proven challenging. The 17 and 19 amino-acid long D4E1 and MsrA3 are among the shortest peptides expressed in plants [41,45]. However, a number of AMPs with relevant properties are shorter. In particular, combinatorial chemistry approaches used to engineer improved synthetic peptides are usually based on smaller lengths [19,20,22,23,47,48]. For expression in plant systems, the length of these peptides should be increased above a minimum threshold while maintaining biological properties [56]. AMPs with high antimicrobial activity have been associated with high toxicity to transgenic plant cells, and peptides with moderate activity have often been expressed, including those especially modified to decrease the antimicrobial activity (e.g. MsrA1, [44]). Foreign AMPs expressed in plants can be prone to cellular degradation by endogenous peptidases, thereby limiting the level of accumulation (see e.g. [57]). Targeting the endoplasmic reticulum (ER) has been suggested as a way to improve accumulation levels, due to either the low proteolytic activity in the lumen or the higher folding ability of ER resident chaperones [58,59]. Confining AMPs to the ER compartment would be expected to decrease their potential toxicity to the plant cell. Moreover, GM rice plants expressing ER targeted cecropin A had resistant phenotypes [29]. Interestingly, any change in the peptide sequence, such as enlargement, the introduction of hinge sequences, ER retention signal, or specific tagging, may affect the biological properties, and this would require a considerable level of screening.
To look at BP100 as a proof-of-concept of the possibility of expressing active AMPs of the BP100 series in plants, a number of BP100 derivatives were rationally designed to be produced in plants. The possible influence of these sequence modifications on the expected antibacterial activity was experimentally tested using chemically synthesized BP100 derivatives and specific bacterial growth inhibition assays. The possibility of producing BP100 derivatives with high antimicrobial activity in a plant system was investigated by stable transformation of rice, following a strategy based on constitutive expression and ER accumulation. The impact of this type of peptide on the fitness of the host plant was specifically evaluated. We envisage our results will be applicable to other short α-helical cationic peptides with high antimicrobial activity.

Results
Rationale of the approach -Design of BP100 derivatives suitable for expression in GM plants BP100 is a synthetic linear polypeptide with only L-amino acids, which is compatible with standard protein synthesis in plant cells. However, BP100 is amidated at the C-terminal position [47]. An unmodified form of BP100 was chemically synthesized, retaining the same amino acid sequence. On analysis, it was found to have similar or slightly lower antibacterial activity and hemolytic activity as BP100 (Table 1), so was selected as a model undecapeptide from the CECMEL11 peptide library for AMP production in plants.
The length of the BP100 undecapeptide, a short molecule, needs to be increased above a minimum threshold to achieve expression in plant systems. Accumulation of this bactericidal peptide in plants could potentially result in BP100 degradation by plant proteolytic activities (e.g. in intercellular spaces) and/or produce plant cell damage even though cytotoxicity and protease susceptibility were minimized during the BP100 selection procedure. This effect is typically diminished in cell compartments such as the ER, where optimal yields of recombinant proteins have often been described. Five BP100 derivative peptides, including up to three copies of BP100 in the same or in opposite orientations, were specifically designed for expression in plants and accumulation in the ER. The derivatives were 15 to 48 amino acids long (see scheme in Figure 1).
It has been previously reported [47] that single amino acid changes in the BP100 sequence have a major effect on its biological characteristics. We tested the antibacterial activity of the newly designed BP100 derivatives by in vitro growth inhibition of the three model plant pathogen bacteria Erwinia amylovora (Ea), Pseudomonas syringae pv. syringae (Pss) and Xanthomonas axonopodis pv. vesicatoria (Xav). The different modifications incorporated in BP100 derivatives designed for expression in plants affected the properties of the resulting peptides (Table 1). BP100.1, BP100.2i and BP100.2mi had similar or higher antibacterial activity compared to BP100, and the increase in activity was higher against Xav than against Ea and Pss. BP100.2mi was the most active peptide and BP100.2 and BP100.3 were the least active BP100 derivatives in this assay.

Toxicity of BP100 derivatives
We initially assessed the possible toxicity to plant producer cells of BP100 derivatives by inoculation of chemically synthesized peptides in tobacco leaves in an in planta assay. Inoculation of 50 to 250 μM BP100.1, BP100.2, BP100.2i, BP100.3 and BP100.2mi induced lesions in tobacco leaves in a dose dependent manner ( Figure 2A), with water as the negative control. The highly cytotoxic peptide mellitin was used as the positive control. BP100 derivatives produced similar or moderately larger lesions than BP100 (Table 1), while mellitin clearly produced more severe lesions at the same doses.
Further assays were carried out with rice. In a germination test, high concentrations of both BP100 and mellitin severely affected seedling development ( Figure 2B). Remarkably, the effects of the two peptides were divergent. Mellitin predominantly inhibited root growth while allowing moderate shoot development. In contrast, BP100 barely effected root growth but almost completely inhibited shoot development. The phenotypic effects of BP100 derivatives were similar to those of BP100. Phytotoxic effects were quantified through shoot lengths of seedlings grown in the presence of decreasing concentrations of peptides, and minimal peptide doses causing significantly shorter seedlings were determined (Table 1). Toxic effects were dose dependent. BP100, unmodified BP100 and BP100.1 produced phenotypic effects at 64 μM and higher concentrations (one-way ANOVA P values at 64 μM were 0.001, 0.001 and 0.000, respectively). BP100.2 produced shorter seedlings at 32 μM and higher concentrations (one-way ANOVA P value at 32 μM, 0.000); and BP100.2i, BP100.3 and BP100.2mi were toxic at 16 μM and higher concentrations (for each peptide at 16 μM, one-way ANOVA P = 0.000) ( Figure 2C). These results demonstrated that BP100 and its derivatives were toxic to plant cells when applied at high doses. The toxicity to mammalian cells of BP100 derivatives was tentatively assessed as the ability to lyse erythrocytes in comparison to a potent hemolytic agent, mellitin. As shown in Table 1, BP100 derivatives generally had higher hemolytic activities than BP100 (except for BP100.1), and this activity seemed to increase with peptide length.

Effect of bp100der transgenes on the efficiency of transformation
We assessed the feasibility of synthetic AMP production in plants using rice as the model, and on the basis of constitutive transgene expression and AMP accumulation in the ER to protect foreign peptides from potential degradation by proteolytic activities and plant cells from phytotoxic activity of peptides. To get a broad picture of the potential and possible limitations of this approach, all five BP100 derivatives were used for transformation, even though they had different characteristics (see above). Chimeric genes encoding all five BP100 derivatives (i.e. bp100.1, bp100.2, bp100.2i, bp100.3 and bp100.2mi, collectively named bp100der) were synthesized and used in a series of transformation experiments designed to quantify the efficiency of the process in terms of amounts of transgenic calluses and plants that could be obtained for each BP100 derivative. A total of 1,200 rice calluses were transformed with each AMP plasmid and the empty vector, with the hptII selection gene and no bp100der sequence, and the numbers of   transgenic calluses were recorded. Regeneration and development of transgenic plantlets were also monitored ( Table 2). Transformation with all five bp100der plasmids resulted in growth of hygromycin-resistant transgenic callus. However, we found major differences in the efficiency of the transformation among plasmids with sequences encoding different BP100 derivatives. Transformation with control plasmids consistently gave the highest numbers of transgenic calluses in our experimental conditions. Slightly lower numbers of calluses showing the selection phenotype were obtained with plasmids carrying bp100.2mi or bp100.2i, while much reduced numbers of transgenic calluses carrying bp100.1 and especially bp100.2 and bp100.3 were identified.
To confirm the functionality of bp100der transgenes in callus with the selection phenotype, we assessed transgene mRNA expression in three randomly chosen transgenic calluses for each construct by RT-qPCR (RNA extraction and cDNA synthesis were systematically performed in duplicate). All analyzed calluses had been subjected to various subculture steps to minimize non-transgenic cells. In a preliminary test we assessed the mRNA levels of three reference genes; elongation factor EF1α, β-actin and 18 S RNA ribosomal genes. In all our callus samples, β-actin had the best score upon application of the GeNorm algorithm (M value = 0.48), so was selected for normalization.
All analyzed calluses expressed both selection marker and (except for those with an empty vector) the corresponding bp100der transgenes (Additional file 1). As expected, transgene expression levels calculated by comparison to β-actin showed great variation among the different events, with relative standard deviations (RSD) of normalized transgene mRNA copy numbers of 63% and 117% for hptII and bp100der, respectively. Compared to β-actin, the amount of selection gene mRNA in GM callus was generally 1-to 10-fold higher, and that of bp100der mRNA 10-to 110-fold higher. The hptII selection gene was similarly expressed in transgenic callus transformed with plasmids carrying different bp100der sequences or the empty vector (5.8 ± 3.7 -fold those of actin; one-way ANOVA P = 0.630). Transgenes encoding distinct BP100 derivatives were expressed at statistically similar levels in calluses exhibiting the selection phenotype (30.3 ± 35.5 -fold those of actin; one-way ANOVA P = 0.283), although, the lowest bp100der expression values consistently corresponded to bp100.2.
Callus with the hptII selection gene and bp100.2 grew slowly, suffered necrosis and never regenerated GM plants. Although they grew normally, none of the~20 bp100.3transformed hygromycin-resistant calluses achieved plant regeneration, so no GM plants expressing this AMP (S-bp100.3) could be obtained with this approach. In contrast, plant regeneration was possible for control, bp100.1, bp100.2i and bp100.2mi transformants (S-hgr, S-bp100.1, S-bp100.2i and S-bp100.2mi plants, respectively), although at decreasing rates. The presence of the bp100der transgene was assessed by qPCR in leaf samples of all regenerated plantlets. Two plants, transformed with bp100.1and bp100.2i-containing plasmids, were qPCR negative and were discarded. An additional plant produced a bp100der qPCR product of different melting temperature than expected. Sequencing of the insert proved the bp100.2mi sequence in this event carried a copy of bp100.2mi with a deletion and was also discarded. Finally, 21, 20 and 8 T0 plantlets were obtained which carried the bp100.1, bp100.2i and bp100.2mi transgenes, respectively ( Table 2). Less than half of these T0 plantlets completed acclimatization and were fertile (although all controls did). The capacity of transgenic cells carrying bp100der sequences to grow as callus, regenerate plants, survive greenhouse standard growth conditions and produce viable seeds was visibly inferior to controls with the empty vector. The efficiency of the whole transformation process was only 10% the control for S-bp100.1, and as low as 2-3% for S-bp100.2i and Absolute numbers and percentages relative to transformation with the control plasmid are given. Control transformation data is given in absolute numbers of hygromycin resistant calluses and percentages of transgenic plants, always relative to the previous step. *A total of 1200 calluses were transformed with each plasmid in three independent experiments, each including transformation with the control vector. Only hygromycin resistant calluses and plants from different calluses (i.e. independent transformation events) were recorded. **Only 15 of the 110 control hygromycin resistant calluses were further processed.
S-bp100.2mi (no S-bp100.2 and S-bp100.3 GM plants were obtained). We concluded that bp100der had a negative effect on transgenic callus and plants; different bp100der sequences had different degrees of phytotoxicity in specific steps of the transformation process.

Molecular and cellular characterization of transgenic plants
A total of fourteen fertile transgenic plants were obtained carrying bp100.2i (2 independent events), bp100.1 (9 independent events) or bp100.2mi (3 independent events). As all T0 transgenic plants originated from independent calluses, they corresponded to different transformation events. For selection of representative GM events, on the basis of insert copy number and bp100der expression levels, genomic DNA extracted from leaf samples of S-bp100.1 and S-bp100.2mi mature T0 plants was analyzed by qPCR targeting bp100der, hptII and actin, to determine approximate transgene copy numbers. Ratios of transgene to actin copy numbers were close to 0.5 for all events, suggesting they all had single copy insertions. Additional leaf samples of the same plants were subjected to RNA extraction and RT-qPCR analysis to compare bp100der mRNA levels. bp100.1 was similarly expressed in the nine different S-bp100.1 events (24.9 ± 5.4 -fold actin mRNA), and bp100.2mi was expressed at similar levels in all three S-bp100.2mi events (2.1 ± 0.8 -fold). Three events carrying bp100.1 (S-bp100.1-6, S-bp100.1-9 and S-bp100.1-10) and two with bp100.2mi (S-bp100.2mi-1 and S-bp100.2mi-9) were selected, in addition to the two events obtained for bp100.2i (S-bp100.2i-5 and S-bp100.2i-42). The seven selected GM events were self-crossed to produce the homozygous T2 generation of GM plants: three independent homozygous lines harboring the bp100.1 transgene and two with each, bp100.2i and bp100.1 were obtained. Southern blot and qPCR analyses confirmed that all had single copies of the transgene (Additional file 2).
Transgene expression was assessed by RT-qPCR in leaves of homozygous plants grown under controlled conditions and sampled at the two-leaf stage. Three biological replicates of ten plants were analyzed per transgenic event. Every transgenic line expressed the corresponding bp100der mRNA in their leaves. Yields of transgene mRNA were calculated by comparison to actin as the reference gene (GeNorm M value < 0.5 in these samples) (Additional file 3). As expected, the expression levels of the transgenes varied in the different events: not only copy number but also the integration site is known to influence transgene expression levels [60].
We then compared transgene expression levels between plants transformed with different bp100der plasmids ( Figure 3). S-bp100.1 and S-bp100.2i plants expressed higher amounts of bp100der mRNA in their leaves than S-bp100.2mi (one-way ANOVA P value, 0.000). hptII mRNA levels were similar in transgenic plants transformed with the different plasmids, including those with the empty vector (one-way ANOVA P = 0.060). The presence of bp100der or the specific bp100der incorporated in GM plants did not affect the expression levels of the selection gene.
The bp100der transgenes encode highly cationic peptides and were designed with ER signal peptide and retention sequences. Accumulation of the bp100der product in the ER could affect the morphology of this organelle. S-bp100.2i.5, S-bp100.1.9, S-bp100.2mi.1 and Senia plants were compared at the ultrastructural level by TEM observation of the crown region of seedlings at the two-leaf developmental stage. Morphology of the ER was clearly altered in S-bp100der cells accompanying vascular cells (Figure 4), with wider and more variable intra-cisternal spaces of ER cisternae and an increase in dictysome vesicles. There was no clear morphological disorganization in organelles such as mitochondria ( Figure 4C) and chloroplasts. An increasing number of vesicles were also formed, particularly in S-bp100.2mi cells, and an accumulation of numerous electrondense granules was detected in parenchyma cells of S-bp100.2mi and S-bp100.1 ( Figure 4D).

Resistance of GM plants to pathogen infection and tolerance to oxidative stress
An indirect approach was used to obtain evidence of the production of BP100 derivative AMPs in our transgenic rice lines, testing their resistance to infection with the bacterial pathogen Dickeya chrysanthemi and the fungus Fusarium verticillioides, and tolerance to oxidative stress. Seeds from all the homozygous lines obtained were incubated in the presence of D. chrysanthemi and F. verticillioides, and germination was monitored. Homozygous plants from all lines were treated with H 2 O 2 in detached-leaf assays. With Senia seeds, infection with 10 2 to 10 6 D. chrysanthemi CFU resulted in progressively smaller, brownish roots and shorter shoots, with germination almost completely inhibited at the highest dose. The pathogenic effect of D. chrysanthemiwas estimated using a semi-quantitative scale ( Figure 5A), and the values recorded seven days after infection ranged from 1.9 (negative control) to 0.4 (10 6 CFU).
We selected 10 5 CFU (with a value of 0.8) to assess susceptibility of S-bp100der seeds to this pathogenic bacterium. S-bp100.1 and S-hgr lines had similar sensitivity values as Senia (one-way ANOVA P values, 0.664 and 0.953, respectively), indicating that hptII or bp100.1 did not significantly modify resistance to this bacterial pathogen. In contrast, S-bp100.2i and S-bp100.2mi lines were more resistant than untransformed plants (one-way ANOVA P values, 0.004 and 0.036, respectively) ( Figure 5B and 5C), signifying that bp100.2i and bp100.2mi transgenes decreased rice susceptibility to D.chrysanthemi.
Seven days after F. verticillioides inoculation of Senia seeds, a cottony mycelium covered the surface of the remaining seed and seedlings showed severe growth effects. The primary embryonic root was shorter than that of uninfected controls, with a brownish tone, and there was little or no crown root production. Shoots were shorter in infected seedlings. The F. verticillioides pathogenic effect was estimated through a semi-quantitative index that reflected shoot height and root development (number and length of crown roots, and length of primary root).
There were clear differences in the effects of this fungus on plants harboring different transgenes. Those transformed with the control plasmid, bp100.2i and bp100.1 had major morphology alterations with susceptibility values similar to Senia (one-way ANOVA P values, 0.562, 0.697 and 0.278, respectively). In contrast, all events harboring bp100.2mi had long shoots and well developed root systems ( Figure 5D), which resulted in higher resistance values (one-way ANOVA P = 0.005). This suggested that rice was protected against infection with F. verticillioides by bp100.2mi transgene but not hptII,bp100.1 or bp100.2i.  F. verticillioides growth in a concentration-dependent manner. The MIC values for the relevant BP100.1, BP100.3 and BP100.2mi peptides ranged from 10 to 40 and <2.5 to 5 μM, respectively (Table 1), which is indicative of their activity against this microorganism. After treatment with H 2 O 2 , Senia untransformed leaves had the widest NBT stained leaf surface (7.3% ± 5.9), indicating accumulation of superoxide radicals (O 2 -). S-hgr rice lines carrying hptII but no bp100der sequences had slightly lower, statistically undistinguishable, NBT staining values (4.0% ± 3.9). NBT staining was somehow lower in S-bp100.1 lines (2.2% ± 2.0), but one-way ANOVA indicated they were similar to S-hgr (P = 0.408), suggesting that there was no clear improvement of stress tolerance with the bp100.1 transgene compared to control leaves. The accumulation of O 2 was considerably lower in S-bp100.2i and S-bp100.2mi rice lines than in untransformed and S-hgr control plants (0.2% ± 0.1 and 0.3% ± 0.7 stained area, respectively; one-way ANOVA P = 0.001; two groups in Tukey's b posttest α = 0.05) (Figure 6), indicating bp100.2i and bp100.2mi transgenes resulted in increased tolerance to oxidative stress.

Performance of GM rice plants with bp100der transgenes
Plants from all S-bp100.2i and S-bp100.2mi homozygous lines obtained in this work were compared with control Senia in terms of agronomic parameters. For each line, three replicates of 10 plants each were analyzed. Growth and yield of control plants were as normal in our culture conditions. No significant differences were found between independent replicates of any particular GM line, and the different lines harboring the same bp100der transgene gave similar values for all studied parameters (one-way ANOVA P values above 0.05).
Slight differences were observed between control plants and those with different transgenes. S-bp100.2i plants were shorter than Senia (observations on 08/25/10), with mean height and standard error values of 90.6 ± 0.9 and 105.3 ± 1.1 cm, respectively (Additional file 4). In contrast, S-bp100.2mi plants were a similar height to the controls (107.8 ± 0.9 cm) (one-way ANOVA P = 0.000, two groups upon Tukey's b posttest with α = 0.05). All lines had the same amount of tillers per plant (7.9 ± 0.2; one-way ANOVA P = 0.854). Leaf chlorophyll content was measured during plant growth and no significant differences

Senia
S-hgr S-bp100.1 S-bp100.2mi S-bp100.2i were observed between S-bp100.2i and Senia lines. S-bp100.2mi exhibited slightly higher values, although they were all in a narrow range (42.8 to 44.6 spad units, observations on 08/25/10) (one-way ANOVA P = 0.012, two groups on Tukey's b posttest with α = 0.05). The Senia control line had better yield, corresponding to the weight of all panicles in a plant, than lines harboring bp100der sequences, and S-bp100.2mi had better yield than S-bp100.2i ( Figure 7A). We refined our yield data by independently considering the number of panicles per plant, number of grains per panicle and grain weight. Senia grains were heavier than those of S-bp100.2mi and S-bp100.2i lines (one-way ANOVA P = 0.05), although overall differences were in a very narrow range (3.6 to 3.8 g/100 grains), as expected. As shown in Figure 7B Senia, S-bp100.2i and S-bp100.2mi had similar numbers of panicles per plant (mean and standard error, 6.2 ± 0.1 panicles/plant), but S-bp100.2mi and especially S-bp100.2i had less grains per panicle, which explained the lower yields.
Complementary experiments were carried out under similar conditions to compare the performance of S-hgr lines (with the hptII selection gene) and untransformed Senia. A total of 5 independent S-hgr events were evaluated. Four were statistically undistinguishable from Senia in terms of all analyzed parameters. One single S-hgr line had similar chlorophyll contents to Senia but it was slightly smaller (99.9 ± 0.6 cm) and had more tillers (9.9 ± 0.4 tillers per plant). There were more panicles, with fewer grains per panicle but of the same weight as in Senia, which gave rise to the same yield as untransformed lines (20.7 ± 0.7 g/plant). These were considered event dependent features.

Discussion
Several synthetic linear undecapeptides of the CECMEL11 peptide library are of increasing interest for future development of fungicides and bactericides against plant pathogens [51]. The linear peptide BP100 is of particular interest due to its properties against major bacterial plant pathogens. Biotechnological production of this short, highly positively charged (pI = 11.02), α-helical amphipathic peptide, which could also damage producing plant cells, is challenging. The aim of this research was to investigate the feasibility of expressing transgenes encoding BP100derived AMPs in transgenic plants using rice as the model host, particularly focusing on the putative impact on the fitness of the host plant. BP100 was used as a model for short cationic AMPs, which are recognized as a remarkable source of active substances with many different applications in diverse sectors besides plant protection [61]. We established the strategy of transgenic peptide accumulation in ER both to minimize the putative toxic effect and to protect it from plant proteases [29,57,62]. The length increase was addressed by designing larger peptides based on the BP100 sequence.
We found that any modification incorporated in BP100 derivatives designed for plant expression altered their properties in terms of antibacterial activity and range of target bacterial species. This is in agreement with previous data on how subtle changes in a peptide sequence of a given length influence antimicrobial activity [19,23,47], and with reports identifying length and sequence among the most relevant factors for biological activity of peptides [63]. It is therefore necessary to assess the biological activity and other relevant properties of any derivative of a given peptide prior to expression in plant systems.
The presence of the C-terminal KDEL element does not result in loss of antibacterial activity (see particularly the comparison between unmodified BP100 and BP100.1), which makes the strategy of peptide accumulation in ER possible. Size increase based on BP100 tandem copies did not result in a proportional improvement of activity, but tended to reduce it. In contrast, peptides with two units in inverted orientation (BP100.2i and BP100.2mi, each unit of the latter elongated with a mellitin fragment) had higher activity than BP100 against the three reporter bacteria. This increase was especially relevant against Xav, demonstrating that sequence variations had an effect on target specificity, which could be related to membrane composition. In view of the antimicrobial activity of these BP100 derivatives, they were considered good candidates to evaluate the possibility of expressing synthetic, highly active antimicrobial peptides in plants.
Monitoring the Agrobacterium based transformation process showed that bp100der transgene products exerted a toxic effect on constitutively producer plant cells, although BP100 has very low toxicity in hemolysis tests [47] and in a mouse model [51]. The complete transformation process of bp100der genes was undoubtedly inefficient, below 10% of control transformations (mean value for all 5 bp100der sequences, 3.2% ± 4.2). Parallel transformation with the empty vector was as expected, so excluding technical errors. Remarkably, no GM plants were obtained carrying bp100.2 or bp100.3, indicating transformation efficiencies of ≤1% that of the empty vector. AMP expression in plants has been accomplished for numerous peptides and host species ( [13,14] and references therein [24][25][26][32][33][34]). Large numbers of GM rice events accumulating Aspergillus giganteus antifungal protein Afp [31] or cecropine A [29] have been reported with normal phenotypes. Recently, thanathin expression has been reported to have very few harmful effects to host plant cells [26]. However, to our knowledge this is the first report on detailed comparison of transformation efficiencies of different putatively toxic transgenes, as the objective is typically to obtain a few transgenic events and the impact of transgene expression at different stages of GM plant production is not addressed.
Not only did transformation with different bp100der sequences have different overall efficiency values but the effectiveness of specific steps during the transformation process varied. As an example, reasonable numbers of calluses harboring bp100.2mi and bp100.2i were obtained (around 80% of the control with only hpII) whereas only around 20 -40% the expected was obtained with bp100.1 and bp100.3. The bp100der transgene mRNA levels in callus could partially explain these results, with somewhat higher expression levels (e.g. bp100.1 and bp100.3) in some way compromising host cell fitness. The number of calluses transformed with bp100.2 was low and they suffered necrosis while expressing the transgene at the lowest bp100der levels. These phenomena can be interpreted on the basis of the different properties of the specific BP100 derivatives: lower levels of bp100.2 product seem to be more toxic to the host cell than higher levels of other bp100der products.
The number of GM calluses regenerating plants was above 60% that of the control for bp100.1 but extremely low for bp100.2mi. In contrast, all regenerated S-bp100.2mi plants but only one fourth of the S-bp100.2i plants survived the acclimatization stage, and virtually all S-bp100.1 plants in the greenhouse produced seeds but less than half S-bp100.2i and S-bp100.2mi developed and were fertile. As the bp100der transgene and mRNA were present in all plants obtained, regeneration of plants mainly from transgenic calluses having lost the bp100der transgenes could be discarded. However, for bp100.2mi only GM plants with low transgene expression levels were viable. Diverse bp100der transgene products seem to have different toxic effects on the producing cells, in particular affecting callus growth and capacity to regenerate plants, plant survival and/or development under standard greenhouse conditions. This is in agreement with the different characteristics of chemically synthesized BP100 derived peptides. Interestingly, length increment based on BP100 tandem repeats did not result in a proportional increase in activity but it greatly increased toxicity to the producer cell. This is in contrast to previous reports associating major antibacterial activity of cationic peptides to higher toxicity to eukaryotic cells [64].
Toxicity of BP100 derivatives to plant cells was confirmed in rice germination and tobacco leaf inoculation assays by comparison of the highly cytotoxic peptide, mellitin. BP100 phytotoxic effects were less intense than those of mellitin and they affected different biological pathways. BP100 reduced shoot development in seedlings while root growth was deficient with mellitin. The five BP100 derivatives had similar phytotoxic effects to BP100 (including BP100.2mi, with part of the mellitin sequence), in some cases with moderately higher intensity. Red blood cell-based assays have been used as an approach to estimate peptide toxicity to mammalian cells [65,66]. The hemolytic activity of the BP100 derived peptides was parallel to phytotoxic activity, with a tendency to increase with peptide length even though mellitin is just 26 amino acids long. Remarkably, toxicity to erythrocytes and plant cells could only be detected upon application of high peptide doses (i.e. 160 to 10-fold target pathogen MIC values). The mechanism of action of AMPs such as BP100 is based on electrostatic attraction and subsequent interaction with cell membranes [50,67]. This property has been exploited by using the BP100 sequence to internalize fused reporter proteins into eukaryotic cells [68]. Differences in membrane lipid composition between bacterial and eukaryotic cells (absence of acidic phospholipids and presence of sterols) drastically reduce susceptibility of eukaryotic cells to many cationic peptides [69] and explains the specific antimicrobial properties of BP100 derivatives. Biologically synthesized bp100der products are expected to accumulate at lower concentrations than those exerting visible phytotoxicity. Phytotoxicity of BP100 derivatives could explain the reduced transformation efficiencies of bp100der transgenes. The bp100der products may cause stress in S-bp100der lines, but its expression certainly can be compatible with moderately decreased fitness and increased pathogen resistance phenotypes (see below).
The possibility of obtaining transgenic rice constitutively expressing bp100der transgenes cannot be directly inferred from the capacity to lyse erythrocytes, damage tobacco leaves upon inoculation or affect seedling development. Transgenic plants were obtained for expression of BP100 derivatives displaying high toxicity values in our assays (BP100.2mi and BP100.2i) but not e.g. BP100.2, with lower values. Exogenous application of peptides at very high concentrations causes these toxic effects, whereas the vulnerability of host plant cells to biologically synthesized bp100der products may differ, as these are produced within the plant cell and are likely to accumulate in a specific cell compartment.
Fertile GM lines with a correctly incorporated and constitutively expressed single copy of the transgene were obtained for some bp100der sequences. The specific chemical properties of BP100 (e.g. high isoelectric point, poor antigenicity, poorly detectable by mass spectrometry) make it extremely difficult to detect and purify from plant cellular material. We are currently working on this, but no operative protocols to directly detect BP100 derived peptides in plant tissues are yet available. In addition, among the prolific number of reports in the current literature on GM plants expressing AMPs only a few gave direct confirmation of AMP production [25,40,41,43]. Nevertheless bp100der mRNA expression in GM plants, low transformation efficiencies and other phenotypic evidences (ultrastructural, oxidative stress and pathogen resistant phenotypes) indirectly demonstrate the production of BP100 derivative molecules in these transgenic plants.
S-bp100der cells exhibit altered ER morphology, with distinct dilation of cisterna and abundant dictysome vesicles. Taking into account that bp100der sequences include signal peptide and ER retention motifs, these observations agree with synthesis of bp100der products and accumulation in this organelle. Additionally, numerous vesicles and electron dense granules were observed in S-bp100.1 and especially S-bp100.2mi cells. Disruption of the ER structure and the presence of increased vesicles or small vacuoles and in some cases, electron dense granules have been associated to exposure of plant cells to excess (toxic) levels of heavy metals [70,71]. This means that these observed morphological characteristics could be related to the cationic nature (as heavy metals) and/or toxic character of BP100 derivatives. Toxicity caused by the expression of bp100.2mi seems to be higher than that of bp100.2i.
Our indicator bacterial species Ea, Xav and Pss are not pathogens of rice so cannot be used to assess resistance of our GM rice plants. However, several antimicrobial proteins in GM plants have been reported to confer protection against a wide range of abiotic and biotic stress conditions. Transgenic expression of the C. annuum antimicrobial protein CaAMP1 in pepper has been shown to confer broad-spectrum resistance against pathogens [72]. In particular, overexpression of cecropin A in rice has been shown to be effective against the fungal blast M. grisea [29] and other conditions such as oxidative stress [73]. Expression of chimerical peptides including cecropin A has been found to give GM potatoes broad-spectrum antimicrobial activity [44], and other cationic α-helical peptides, such as the MsrA2 derivative of Dermaseptin B1 from Phyllomedusa bicolor, have been reported to confer resistance to a variety of fungal and bacterial phytopathogens in transgenic potato plants [42].
Activity of chemically synthesized BP100, BP100.1, BP100.2i and BP100.2mi peptides against the soft rot pathogen D. chrysanthemi (syn. Erwinia chrysanthemi [74]) was dose-dependent. Although they are mainly antibacterial, BP100 derivatives additionally exhibited antifungal activity against F. verticillioides, associated with the bakanae disease of rice [75]. This correlated well with the increased resistance of S-bp100.2mi lines to F. verticillioides and S-bp100.2mi and S-bp100.2i lines to D. chrysanthemi. Biotic and abiotic stresses are typically associated with the rapid production of reactive oxygen species (ROS), including H 2 O 2 and O 2 -. ROS are known to play dual roles; they play a central role in the regulation of biological processes such as stress response, hormone signaling and development, but high ROS levels have been implicated in the damaging effects of various environmental stresses [76]. S-bp100.2mi and S-bp100.2i vegetative tissues subjected to oxidative stress showed decreased accumulation of O 2 radicals, suggesting an increased ability to scavenge ROS. Overexpression of bp100.2mi and bp100.2i led to enhanced resistance to bacterial and fungal pathogens; and improved tolerance to oxidative stress.
The resistant phenotype of S-bp100.2mi and S-bp100.2i was not solely an outcome of the selection gene and/or the transformation itself as shown by (i) the different susceptibility phenotypes exhibited by S-bp100.2i and S-bp100.2mi lines; and (ii) the lack of resistance of untransformed Senia and the GM S-hgr and, unexpectedly, S-bp100.1 lines. Chemically synthesized BP100.1 was active against our indicator bacterial species, D. chrysanthemi and F. verticillioides, but bp100.1 was the less phytotoxic bp100der transgene. Taking into account its extremely small length (15 amino acids), we can speculate that this BP100 derivative was rapidly degraded in S-bp100.1 plants and/or produced in a modified form (e.g. not properly processed).
The expression of foreign genes in plants can trigger the activation of plant defense mechanisms normally activated only during pathogenesis [77]. Several resistant GM rice lines such as those constitutively expressing cecropin A [73], the antifungal protein AFP [78] or a pathogenesis-related (PR) protein [79] have been reported to overexpress endogenous defense genes in the absence of the pathogen. This means we cannot rule out transgene dependent overexpression of stress genes being the cause of the resistance phenotypes observed in S-bp100.2i and S-bp100.2mi lines. However, reverse transcription coupled to real-time PCR analyses of PR1b and PR5 coding genes [GenBank: U89895, X68197, widely used indicators of induction of plant defense responses [80]] showed they were expressed at similar levels in leaves of S-bp100.2i and S-bp100.2mi GM when compared to Senia in vitro grown plants. Additionally, plants overexpressing transgenes encoding ER driven proteins have been reported to constitutively express the unfolded protein response (UPR), activated by misfolded protein accumulation in the ER [73]. The UPR has recently been associated to plant resistance to abiotic stress and pathogen attacks ( [81], and references therein), so the accumulation of transgene products (expected to be highly cationic peptides) in ER could result in increased-resistance phenotypes. In a preliminary experiment we showed that S-bp100.2mi and S-bp100.2i were somewhat resistant to the causative agent of rice blast M. grisea, while we could not detect clear in vitro activity of the corresponding peptides against M. grisea spores (data not shown). S-bp100.2i and S-bp100.2mi pathogen resistant and oxidative stress-tolerant phenotypes support the expression of active antimicrobial peptides in these plants, irrespective of whether these phenotypes are peptide-direct effects or indirectly derived from transgene dependent activation of other cellular pathways. The latter seems to have a role in S-bp100.2i decreased susceptibility to F. verticillioides, taking into account that BP100.2mi and BP100.2i have similar MIC values against this pathogen, whereas evidence from ultrastructural analysis and mRNA expression suggest the bp100.2mi product exerts higher toxicity on host cells than the bp100.2i product.
Although with low efficiency, we showed the feasibility of producing rice lines expressing at least some bp100der sequences (e.g. S-bp100.2i and S-bp100.2mi) and indirectly demonstrated the production of BP100 derivative molecules. Thorough assessment of agronomic characteristics showed the nutritional status of these GM plants was unaffected by bp100der transgenes, as demonstrated by leaf chlorophyll content: chlorophyll content has been directly related to nitrogen content, which is considered an indicator of plant nutritional status. Although S-bp100.2i plants were around 10% shorter than expected, the greatest reduction in the measured parameters was the number of grains per panicle in GM (especially S-bp100.2i) compared to control lines. It is well known that stress conditions have an effect on the numbers of grains per panicle, so it could be speculated that this was a consequence of the phytotoxic effect of BP100 derivative accumulation in plant cells. The slight reduction in grain weight is in agreement with it being a well conserved character under different cultural conditions. Despite this, the overall performance of rice lines expressing certain bp100der transgenes strongly resembled comparable untransformed lines. The best case was S-bp100.2mi, with similar vegetative features and only around 10% grain yield losses compared to Senia.

Conclusions
We found that transformation of rice with genes encoding α-helical cationic AMPs such as those derived from the synthetic 11 amino acids-long BP100 had, in many cases, a highly negative effect on the efficiency of transformation. Remarkably, GM lines constitutively expressing certain bp100der sequences were produced with resistance phenotypes and minimal impact on agronomic performance. BP100 tandem copies could not be constitutively expressed in GM plants, but sequences encoding BP100 inverted repeats either elongated or not with portions of a natural AMP (bp100.2mi and bp100.2i) greatly diminished transgene undesired effects. It is likely that GM plants constitutively expressing bp100der transgenes and displaying resistance to representative plant pathogens and fit phenotypes are the result of a complex equilibrium between bp100der product phytotoxicity, antimicrobial activity and transgenedependent plant stress response. This approach could be considered feasible for expressing synthetic AMPs in plants either to establish disease resistant plants or to facilitate molecular farming.

Methods
Chemical synthesis of BP100 derived antimicrobial peptides BP100 derivative peptides were designed with increasing lengths and inclusion of an ER retention signal (KDEL four-amino acid sequence). They included two or three copies of BP100, with a four amino-acid linker element, AGPA, which structurally links the α-helical active domain and the C-terminal domain in natural cecropin A [82]. One of the designed peptides was further elongated by adding a portion of mellitin (amino acids 10 to 18), another another natural AMP. Two BP100 derivatives were designed with monomeric units in opposite orientations (see scheme in Figure 1). BP100 derived peptides were manually synthesized by the solid-phase method using Fmoc-type chemistry as previously described [47]. Briefly, Fmoc-Rink-MBHA resin (0.64 mmol/g) was used as solid support, and sidechain protection was performed with tert-butyloxyrocarbonyl for Lys and Trp, and tert-butyl for Tyr. HBTU and DIEA (3 equiv. each) mediated couplings of Fmoc-amino acids (3 equiv.) in N,N-dimethylformamide (DMF), monitored by the ninhydrin test. After removal of the Fmoc group with piperidine-DMF (3:7) the peptidyl resin was washed with DMF. On completion of the sequence, peptides were cleaved from the resin with TFA-H 2 O-TIS (95:2.5:2.5) and dissolved in H 2 O and lyophilized after TFA evaporation and diethyl ether extraction. HPLC was used to assess their purity, which was above 90% in all cases. Peptide identity was finally confirmed by electrospray ionization mass spectrometry.
In vitro assessment of the properties of BP100 derivatives Serial dilutions of synthetic peptides were prepared at 400, 200, 100, 50, 25, 12.5 and 6 μM. Twenty μl aliquots were mixed in a microtiter plate with 20 μl bacterial suspensions (10 8 CFU/ml) and 160 μl trypticase soy broth (TSB, BioMèrieux, France), or 80 μl fungal suspensions (10 4 conidia/ml) and 100 μl of double-concentrated PDB containing 0.006% (w/v) chloramphenicol. Three replicates were carried out per microbial strain, peptide and concentration. In the experiments, water was used as the positive control instead of the peptide, or as the negative control instead of the microbial suspension. Microplates were incubated at 25°C for 48 h (bacterial strains) or 22°C for 6 days (Fv), and optical densities at 600 nm were recorded hourly (every two hours for Fv) after 20 s shaking. Two experimental replicates were performed. The lowest peptide concentration showing negative microbial growth at the end of the experiment was taken as the minimal inhibitory concentration (MIC).

Phytotoxic activity
For each treatment, a total of 12 surface-sterilized Senia seeds were germinated in 500 μl water or peptide solution at the appropriate concentration (0, 8, 16, 32 and 64 μM) in a culture chamber (28 ± 1°C with a photoperiod of 16 h light / 8 h dark under fluorescent Sylvania Cool White lamps) for seven days. Seedling morphology was recorded and shoot length was measured. Mellitin and BP100 were also tested at 125 and 250 μM.
Tobacco (Nicotiana benthamiana) plants were grown from seed in a heated glasshouse and used between 20 and 30 days old. One hundredμl chemically synthesized peptides at 50, 100, 150 or 250 μM were inoculated into the mesophylls of fully expanded tobacco leaves (previously wounded with a needle), using a syringe without needle, and plants were kept at standard glasshouse conditions for three days. Up to six independent inoculations were carried out in a single leaf, and at least three independent inoculations were performed per peptide and concentration, randomly distributed in different leaves and plants. Toxicity was measured as the lesion diameter.

Hemolytic activity
Hemoglobin release from erythrocyte suspensions of fresh human blood (5% v/v) was used to evaluate hemolytic activity of synthetic peptides [47]. Briefly, aseptically collected blood (BD vacutainer K2E System with EDTA, Belliver Industrial State, Plymouth, UK) was centrifuged at 6,000 g for 5 min, washed with 10 mM TRIS, 150 mM NaCl, pH 7.2 and suspended in the same buffer. Fifty μl of peptide solutions at 300 μM were mixed with 50 μl aliquots of erythrocyte suspensions and incubated for 1h at 37ºC with shaking. After centrifugation, the supernatants were transferred to microplate wells with 80 μl H 2 O and the 540 nm absorbance was monitored with a Bioscreen plate reader. Hemolysis percentages were calculated relative to mellitin and TRIS buffer.

Construction of the chimeric bp100der genes and plant expression vectors
A wide variety of sequence-dependent factors influence gene expression levels and we took several critical ones into account. The deduced DNA sequences encoding BP100 derivatives were codon-optimized for more preferred codon usage in rice plants [83]. The distribution of codon usage frequency along the sequence was good, with codon adaptation index (CAI) values from 0.82 to 0.87 (CAI > 0.8 indicates a high level of expression). All five sequences had average GC contents close to 50% and no peaks outside the 30-70% range were found in 60 bp windows, which could reduce mRNA half-life. No internal ribosomal binding sites or other putatively destabilizing cis-acting elements were found. The DNA sequences encoding BP100 derivatives are shown in Additional file 5.
The signal sequence from the tobacco pathogenesis related protein PR1a [84] was fused to each AMP coding sequence to drive the BP100 derivative peptides to the secretory pathway. With the KDEL C-terminal sequence, it has been successfully used to drive foreign proteins to the ER in rice var. Senia [29]. Chimerical genes encoding BP100.1, BP100.2 and BP100.3 (bp100.1, bp100.2 and bp100.3 genes) were fully synthesized by recursive PCR [85] in a single reaction. Four oligonucleotides (i.e. two in sense and two in antisense orientations, Table 3 and Figure 1) were designed to cover the whole bp100.3 sequence with partial overlapping. The outermost oligonucleotides incorporated BamHI restriction sites for subsequent cloning. They were synthesized on an Applied Biosystems 394 DNA synthesizer according to [31].
Recursive PCR was carried out in a final volume of 100 μl with 30 pmol external and 0.2 pmol internal primers, 300 μM dNTPs, 1x reaction buffer, 1.5 mM MgCl 2 and 1.75 units Expand High Fidelity DNA polymerase (Roche Diagnostics Corporation, Indianapolis USA). Reaction conditions were as follows: initial denaturing step (3 min 94°C); 10 cycles of 15 s at 94°C, 30 s at 45°C and 3.5 min 72°C; 20 cycles of 15 s at 94°C, 30 s at 58°C and 3.5 min 72°C; and a final elongation step of 10 min at 72°C. Due to the repetitive nature of bp100.3, this amplification consistently resulted in a mixture of three bands with lengths compatible with sequences containing one, two and three tandemly repeated bp100 units. Recursive PCR products were cloned using the pGEM W T-Easy system (Promega, Wisconsin, USA) according to the manufacturer's instructions. Individual clones were subjected to PCR with primers PR1a_for and KDELbam_rev (Table 3) to individually amplify bp100.1, bp100.2 and bp100.3. The PCR was in a final volume of 50 μl 1x reaction buffer with 1.5 mM MgCl 2 , 200 nM primers, 200 μM dNTPs and 1 unit Expand High Fidelity DNA polymerase (Roche Diagnostics Corporation). Reaction conditions were 3 min at 95°C; 40 cycles of 30 s at 95°C, 30 s at 58°C and 30 s at 72°C; and 10 min at 72°C. Synthetic bp100.2i and bp100.2mi chimerical genes, encoding BP100.2i and BP100.2mi, were purchased from GenScript (Piscataway NJ, USA) and included as well BamHI restriction sites, both at the 5' and 3' ends.
The newly synthesized bp100.1, bp100.2, bp100.2i, bp100.3 and bp100.2mi genes were subcloned into the BamHI site of pAHC17 plasmid DNA [86], flanked by the promoter, first exon and first intron of the maize ubiquitin gene [87], previously shown to drive high expression levels of transgenes in rice [86,88], and the Agrobacterium tumefaciens nopaline synthase nos terminator sequences. After sequence verification by sequencing, the complete bp100der sequences (including promoter and terminator elements) were subcloned into the KpnI site of pCAMBIA1300, always in the opposite sense compared to the selection gene. The pCAMBIA derived binary vectors harboring bp100der cassettes were transferred into Agrobacterium strain EHA105 using the cold shock method [89]. DNA manipulations and transformation of the Escherichia coli strain XL1Blue with plasmid DNA were carried out using standard techniques [89].

Production of transgenic rice plants
Transformation was carried out using the Mediterranean elite japonica rice (Oryza sativa L.) cultivar Senia. Transgenic lines expressing one of the bp100der genes were produced by Agrobacterium mediated transformation of embryonic callus derived from mature embryos as described by [90]. The control plasmid was pCAMBIA 1300 (with the resistance to hygromycin selection gene). T0 plants were grown to maturity and a selection of events was further cultured in standard greenhouse conditions to obtain homozygous transgenic lines in the T2 generation.

Nucleic acid extraction
Genomic DNA was extracted from 1 g plant material using a CTAB based method [91]. Total RNA was extracted from 400 mg plant material, using a protocol based on the Trizol reagent (Invitrogen Life Technologies, Carlsbad, CA, USA), and purified using the Qiagen RNeasy MiniElute Cleanup kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. DNA and RNA concentration and quality were systematically checked by UV absorption at 260 and 280 nm using a NanoDrop ND1000 spectrophotometer (Nanodrop technologies, Wilmington, DE, USA). All samples had appropriate values (mean and standard deviation [SD], 1.94 ± 0.15 and 2.08 ± 0.02 for DNA and RNA extracts, respectively).

Southern blot
Genomic DNA (25 μg) was digested with Hind III or Eco RI, electrophoresed on 0.8% agarose gels, transferred to nylon membranes (Hybond-N, GE Healthcare Life Sciences, UK) and hybridized with a thermostable alkaline phosphatase labeled probe. The bp100der sequences were detected by chemiluminescence using the CDP-Star™ reagent (GE Healthcare Life Sciences, UK). The bp100der probe was obtained from a 406 bp PCR product encompassing part of the ubi first intron, the pr1a signal sequence and the first bp134 element (PCR amplified with primers SouthUBI_for and SouthBP_rev, Table 3 and Figure 1) using the AlkPhos Direct Labeling and Detection System (GE Healthcare Life Sciences, UK). Hybridization and washes were carried out at 65°C according to the manufacturer's instructions.

Reverse transcription and PCR analysis
Transgene expression was assayed by reverse transcription coupled to real-time polymerase chain reaction (RT-qPCR). Reverse transcription was performed on 2,000 ng total RNA, previously treated with Turbo DNase (Ambion, Austin, TX, USA) using 50U of Multi-Scribe Reverse Transcriptase (Applied Biosystems, Foster City, CA, USA) and random hexamer primers (High-Capacity cDNA Reverse Transcription Kit, Applied Biosystems, Foster City, CA, USA) according to the manufacturer's protocol. For each sample, cDNA was prepared at least in duplicate. The absence of remaining DNA targets was demonstrated by qPCR analyses (see below) of DNase-treated RNA samples.
qPCR assays targeting all five bp100der sequences and hptII were developed based on SYBR-Green technology. A single bp100der RT-qPCR assay was designed and optimized to detect and quantify bp100.1, bp100.2, bp100.2i, bp100.3 and bp100.2mi based on two primers targeting conserved sites at the 5' and 3' ends of the coding sequences. PCR primers were designed using the Beacon Designer 7.0 software (Premier Biosoft International, Palo Alto, CA, USA). The qPCR assays were in a 20 μl volume containing 1X SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA), 100 nM of primers (except for hgr_for and hgr_rev, 300nM and bp100der_rev, 50 nM) and 1 μl cDNA. Reaction conditions were: (1) initial denaturation (10 min at 95°C); (2) amplification and quantification (50 repeats of 15 s at 95°C and 1 min at 60°C) and (3) melting curve program Table 3 Oligonucleotides used in this work Underlined sequences indicate overlaps between two oligonucleotides used in recursive PCR, in italics is the pr1a signal sequence, the sequence encoding the KDEL retention signal is indicated in bold, and the BamHI and KpnI restriction sites introduced for cloning purposes are shown in lower case. Length of oligonucleotides is indicated in nt.
(60-95°C with a heating rate of 0.5°C/s). Melting curve analyses produced single peaks, with no primer-dimer peaks or artifacts, indicating the reactions were specific. All oligonucleotides (Table 3) were purchased from MWG Biotech AG (Germany). Reactions were run on a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) in triplicate. All reactions had linearity coefficient (R 2 ) and efficiency values [E = 10 [−1/slope] , [92]] above 0.99 and 0.95, respectively (for bp100der qPCR these values were assessed for each target). The 18 S ribosomal RNA, β-actin and elongation factor (EF1α) housekeeping genes were analyzed as previously described [78]. Their suitability as internal standards was assessed in our samples through the geNORM v3.4 statistical algorithm, and β-actin had M values below 0.5 in all cases.

Pathogen infection assays and oxidative stress assay
A total of ten homozygous plants of each GM event (3 S-bp100.1, 2 S-bp100.2i and 2 S-bp100.2mi lines), plus two control lines (i.e. untransformed Senia and S-hgr, transformed with the empty plasmid) were grown in the greenhouse to the vegetative three-leaf stage. The second leaves were detached and 50 mM H 2 O 2 applied for 8 h, with subsequent in situ detection of O 2 radicals in plant tissues by overnight staining with nitro blue tetrazolium (NBT). Leaves were visually inspected and the percentages of stained area were calculated using the APS assess v2.0 software tool.
A total of 15 homozygous seeds of each GM event, plus Senia and S-hgr lines, were surface sterilized and germinated in sterile water in 24-well culture chambers. They were incubated in a culture chamber at 28 ± 1°C with a photoperiod of 16 h light / 8 h dark under fluorescent Sylvania Cool White lamps. After overnight pre-germination they were inoculated with increasing concentrations of the bacterial plant pathogen D. chrysanthemi (10 2 , 10 3 , 10 4 , 10 5 and 10 6 for Senia; 10 5 for all GM events) and allowed to continue germination for seven days. The development of D. chrysanthemi-infected seedlings from wild-type and transgenic plants was determined using a semi-quantitative scale ( Figure 5A). Control seeds incubated in water confirmed germination of over 98% seeds, with no statistical difference among different events (one-way ANOVA p = 0.617). Uninfected seedlings from all analyzed events consistently had maximum development values.
For F. verticillioides (anamorph stage of Gibberella fujikuroi, mating population A) assays, nine seeds of each GM event, plus Senia and S-hgr lines, were surface sterilized and incubated for 30 min in the presence of a suspension of F. verticillioides conidia (10 5 spores/ml) under agitation. They were allowed to germinate for seven days in sterile MS medium (Murashige and Skoog, 1962) supplemented with 0.7% agar in plastic tins under the same conditions as for D.chrysanthemiassays. Seedling development was estimated through a semiquantitative index that considered shoot height (with values in the 0 to 1 range) and root development (number and length of crown roots and length of primary root, with values of 0 to 0.75 and 0 to 0.25, respectively). As uninfected control seeds consistently germinated and seedlings developed as expected they were given the highest possible values in this assay (i.e. 1, 0.75 and 0.25, respectively). Shoot height and primary root length were measured and values were assigned in the 0 to 1 and 0 to 0.25 ranges, respectively, in comparison to control seedlings. The number and length of crown roots were estimated by comparison with crown roots in control seedlings, and values were in the 0 to 0.5 and 0 to 0.25 ranges, respectively. The sum of the values for all seedlings was used to calculate the value of the seedling development index for each GM event. A global value of 0 indicated no germination and compact mycelium over the seed, while a value of 2 represented standard growth, the same as uninfected seeds.

Agronomic characterization of GM plants
Transgenic and control lines were compared in terms of agronomic parameters. For S-bp100.2i and S-bp100.2mi lines (i.e. those expressing bp100.2i and bp100.2mi), T3 homozygous plants from all the independent events obtained were assayed. The assay was carried out in the quarantine greenhouse from 29 April to 7 October 2010, i.e. during the conventional rice growing season in the region. For each line, a total of 45 seeds were sown and 81 to 91% germinated. At the vegetative three-leaf stage a total of 30 plants per line (three replicates of 10 plants each) were transferred to pots and grown in the greenhouse under standard conditions. Plant growth was monitored every two weeks in terms of plant height, number of tillers, chlorophyll contents (spat) and flowering (a total of 5 observations, the last on 08/25/2010). To estimate yield related traits, the number of panicles per plant, number of grains per panicle and the weight per 100 grains were measured on harvesting. Plant yield was calculated as the product of number of panicles, number of grains per panicle and grain weight, and corresponds to the weight of all panicles in a plant.

Bioinformatics
RT-qPCR data were normalized with the housekeeping gene and statistically analyzed using the SPSS software v.15.0 for Windows or the Genex software v.5.1.1.2 (MultiDAnalyses).
For pathogen resistance and oxidative stress tolerance assays, one-way ANOVA analyses were performed. The analytical results (i.e. percentages of NBT stained area or seedling development values in the presence of D. chrysanthemi or F. verticillioides) were used as the dependent variables, and the transgene (bp100.2i, bp100.1, bp100.2mi, nptII or none) as factor. Statistical analyses were performed jointly comparing the different events carrying the same bp100der to Senia.