Several synthetic linear undecapeptides of the CECMEL11 peptide library are of increasing interest for future development of fungicides and bactericides against plant pathogens
. 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 BP100-derived 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
. 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
. 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
 and in a mouse model
. 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
[24–26, 32–34]). Large numbers of GM rice events accumulating Aspergillus giganteus antifungal protein Afp
 or cecropine A
 have been reported with normal phenotypes. Recently, thanathin expression has been reported to have very few harmful effects to host plant cells
. 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
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
. 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
 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
. In particular, overexpression of cecropin A in rice has been shown to be effective against the fungal blast M. grisea and other conditions such as oxidative stress
. Expression of chimerical peptides including cecropin A has been found to give GM potatoes broad-spectrum antimicrobial activity
, 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
Activity of chemically synthesized BP100, BP100.1, BP100.2i and BP100.2mi peptides against the soft rot pathogen D. chrysanthemi (syn. Erwinia chrysanthemi) was dose-dependent. Although they are mainly antibacterial, BP100 derivatives additionally exhibited antifungal activity against F. verticillioides, associated with the bakanae disease of rice
. 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 H2O2 and O2
-. 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
. S-bp100.2mi and S-bp100.2i vegetative tissues subjected to oxidative stress showed decreased accumulation of O2
- 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
. Several resistant GM rice lines such as those constitutively expressing cecropin A
, the antifungal protein AFP
 or a pathogenesis-related (PR) protein
 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
] 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
. The UPR has recently been associated to plant resistance to abiotic stress and pathogen attacks (
, 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.