Introduction

Although most animal and plant defensins kill or inhibit the growth of microbial pathogens directly (Schroeder 1999), some of these cationic peptides have been shown to activate the adaptive immune system (Yang et al. 1999) or mediate developmental events (Zhou et al. 2004). An epididymis-specific β-defensin proved to be important for the acquisition of sperm motility and initiation of sperm maturation (Zhou et al. 2004). Interaction between a defensin-like peptide secreted from pollen with its cognate S-locus receptor kinase on the stigma triggers a self incompatibility response to prevent self fertilization in Brassica (Takayama et al. 2001; Yang et al. 1999).

More than 300 defensin-like peptides exist in the Arabidopsis genome, 78% of which have a cysteine-stabilized α-helix β-sheet (CSαβ) motif common to plant and invertebrate defensins (Silverstein et al. 2005; Thomma et al. 2002). Defensin-like genes have been grouped into classical, seed- and nodule-specific defensins based on the number and positions of cysteine residues (Graham et al. 2008). Analysis of 1,100 expressed genes from 26 different species indicated that 60% of the defensin-like subgroups are restricted to single taxonomic clades. Taxonomic expansion was most striking in the case of nodule-specific defensin-like genes from legumes. It has been speculated that an originally antimicrobial peptide was co-opted to function as a defensin-like peptide in self incompatibility (Nasrallah 2002). Additional functions in plant development are likely because defensins have been shown to inhibit the growth of roots and root hairs in Arabidopsis thaliana (Allan et al. 2008).

Like other plants, Solanaceae produce archetypal defensins [Supplemental Fig. S1A]. A subset of these defensins contains an acidic carboxyterminal domain following the mature peptide domain (Supplemental Fig. S1B). A prodefensin from tobacco has previously been shown to undergo processing (Lay et al. 2003), but the functional significance of removal of the C-terminal prodomain remains unknown. By comparison, processing is required for antimicrobial activity of mammalian α-defensins (Wilson et al. 1999). Most plant defensins are active against fungi (Thevissen et al. 2000) whereas others inhibit α-amylase activity from insects (Bloch and Richardson 1991) or protein synthesis (Mendez et al. 1990). Defensins are expressed in tissue-specific patterns (Ferrandon et al. 1998; Thomma and Broekaert 1998). In a study of flower development, a defensin from tomato was identified as one of six pistil-expressed genes (Milligan and Gasser 1995). This defensin, here referred to as DEF1 (Supplemental Fig. S1B), is expressed in the outermost layers of leaf primordia and floral organs (Brandstadter et al. 1996). We report herein functional analysis of the closely related DEF2 gene by altering its expression in transgenic tomato plants.

Materials and methods

Plant material and growth conditions

Seeds of tomato (Solanum lycopersicon) cv. VF36 and cv. Zhongshu 5 were obtained from the Tomato Genetics Resource Center and Les Fuchigami (Wang et al. 2005), respectively. Plants were grown in 2 l pots containing Sunshine SB40 soil mix (SunGro Horticulture, Bellevue, WA) supplemented with osmocote fertilizer (15-9-12) under greenhouse conditions with supplemental lighting using high-pressure sodium lamps (400 W). Day and night time temperatures were 21 and 16°C, respectively.

Construction of bacterial expression vectors

Defensin clones TPP3 and cTOB11C9 were obtained from Charles Gasser (Department of Molecular and Cellular Biology, University of California, Davis) and Clemson University Genomics Institute, respectively. The pMAL-c2X vector (New England Biolabs, Beverly, MA) was used for bacterial expression of defensins. Primers 5′-CGGAATTCGGATCCCAAATTTGCAAAGCACCAAGC-3′ and 5′-GCTCTAGAGGATCCTTAACATGGCTTAGTGCATAGACACTTC-3′ were used for amplification of DEF1 with Accuzyme (Bioline, Springfield, NY); primers 5′-CGGAATTCGGATCCCAGATGTGCAAATCAACAAGC-3′ and 5′-GCTCTAGAGGATCCTTAACAAACCTTAGTACATAGGCACTTTC-3′ were used for amplification of DEF2. The annealing temperature was 65°C. PCR products were directionally cloned into pMAL-c2X using EcoRI and XbaI and transformed into E. coli BL21. DNA inserts were sequenced. IPTG was used to induce expression. Amylose resin and maltose were used for protein purification and elution. Recombinant DEF1 and DEF2 peptides were cleaved with Factor Xa to release mature defensins.

Construction of plant expression vectors

pBinVec3 was obtained from Jeff Leonhard (Crop and Soil Science, Oregon State University, Corvallis). pBinVec3 is a derivative of pGPTV-KAN (Becker et al. 1992). The uidA::pAnos cassette of pGPTV-KAN was replaced with the pAg7::pmas::pAocs and pCaMV35S::pA35S cassettes of pPCV91 using EcoRI and HindIII sites (Martin et al. 2001; Strizhov et al. 1996). The resulting orientation of transcription from the quadruple CaMV 35S promoter is opposite to the direction of the nptII gene. Primers 5′-CGGGATCCCATGGCTCGTTCCATTTGCTTC-3′ and 5′-CGGGATCCTTACTCCATCACAATCTCTTCTTCAAG-3′ were used to amplify the complete open reading frame of DEF2 with Accuzyme. The annealing temperature was 65°C. DEF2 was inserted between the quadruple CaMV 35S promoter and the 35S terminator using a unique BamHI cloning site. XcmI was used to determine insert orientation. Plasmids containing DEF2 in the sense or antisense orientation were transformed into the Agrobacterium tumefaciens EHA105. The DNA insert was sequenced prior to plant transformation.

Plant transformation

Tomato cv. Zhongshu 5 was transformed according to a published protocol (Wang et al. 2005), except that 500 mg l−1 cefotaxime was used to kill Agrobacterium after co-cultivation. Plantlets were transferred to soil, acclimated, and moved to the greenhouse. Transgenic T0 lines were self-pollinated. Subsequent T1 and T2 generations were selected on ½ MS, 15 g l−1 sucrose, 100 mg l−1 kanamycin.

Analysis of plant genomic DNA

Genomic DNA was extracted from tomato leaves using cetyltrimethylammonium bromide (Fulton et al. 1995). For Southern hybridization (Sambrook and Russell 2001), DEF2 insert was excised with EcoRI and XhoI and used in conjunction with an AlkPhos Direct Labeling Kit (Amersham Biosciences, Piscataway, NJ) for detection of hybridizing bands.

The CaMV 35S terminator-specific primer 5′-TCTTATATGCTCAACACATGAGCG-3′ was used in combination with a DEF2-specific forward 5′-ATCAACAAGCCAAACCTTCAAG-3′ or reverse primer 5′-GCTTCCCCAACCAAAGTTGT-3′ to determine the orientation of sense and antisense constructs, respectively. Taq polymerase (GeneScript, Piscataway, NJ) was used for amplification with an annealing temperature of 55°C.

Reverse transcriptase (RT)-PCR and real-time PCR

Total RNA was extracted from leaves or floral buds and organs using the phenol-LiCl method (Sambrook and Russell 2001). cDNA was synthesized using 2 μg of RNA and a first-strand synthesis kit (GE Healthcare, Piscataway, NJ). Endogenous gene expression was traced using the forward primer for DEF2 in combination with a primer 5′-AAAGCCCAATAACACGACATT-3′ specific to the 3’UTR of DEF2. Transgene expression was monitored using forward or reverse DEF2 primer in combination with the CaMV 35S terminator-specific primer. Primers 5′-CGAGAGAAGATGACTCAGATC-3′ and 5′-AGGTGGGGCGACAACCTTG-3′ were used to amplify an intron-spanning piece of the actin Tom52 (Moniz and Drouin 1996). Taq polymerase was used for amplification with an annealing temperature of 55°C. DEF1 was amplified using primers 5′-CTCTTTGTTACCTATGAGGTAG-3′ and 5′-CTTCCTCACCCAAAGTTGCT-3′.

Primers and probes for real-time PCR of DEF2 were 5′-GACCATGGCTCGTTCCATTT-3′, 5′-CAAAGAGCACCATTGCCAAGA-3′, and 5′-6FAM-CCTCATGGCACTTATG-MGBNFQ-3′. Primers and probes for the endogenous DEF2 were 5′-GCTTGAAGAAGAGATTGTGATGGA-3′, 5′-ACACGACATTAGAAGCTACCCTTTTAC-3′, and 5′-6FAM-TTGAGTGTCAAAATCA-MGBNFQ-3′. A VIC/MGB-labeled eukaryotic 18S endogenous control (Applied Biosystems, Foster City, CA) was used to determine relative mRNA abundance. Triplicate samples were analyzed. DEF2 plasmid and salmon sperm DNA were used for calibration using the Relative Standard Curve Method, as recommended by the manufacturer.

Immunoblot analysis

A peptide-specific polyclonal antibody was generated against the surface-exposed defensin epitope QTFKGLaFTD, located between the first β-strand and the α-helix. An Ala was inserted at position 7 because the Cys residue is likely engaged in a disulfide bond and not accessible to the antibody. The peptide antigen and the antibody from rabbit were commercially generated (Biocarta, San Diego, CA).

Proteins were extracted in 130 mM acetic acid and subsequently incubated for 2 h on ice (Chipps et al. 2005; Schröder et al. 1992). An equal volume of ethanol was added to the extract and the mixture was stored in the freezer overnight. After centrifugation at 10,000g for 10 min, the pellet was resuspended in sample buffer (Laemmli 1970). The supernatant was concentrated in a Savant SpeedVac Concentrator (Thermo Scientific, Waltham, MA) and resuspended in 0.01% (v/v) acetic acid. Soluble proteins were quantified using the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA). Proteins extracted in sample buffer were quantified using the RC DC Protein assay (Bio-Rad, Hercules, CA).

Peptides were separated using a modified Tricine-based SDS-PAGE method (Schagger and von Jagow 1987; Titus 1990) and transferred to ProBlott membrane (Applied Biosystems, Foster City, CA). An ECL Western Blotting Analysis System (GE Healthcare, Piscataway, NJ) was used for antigen detection.

Phenotypic analysis

Plant height and fresh weight of ripening tomato fruits were measured. Numbers of seeds per fruit (n ≥ 3) were counted and the age at which mature fruits were harvested was determined.

Plant inoculation

Foliar inoculation assays were performed as published (Guimaraes et al. 2004). Detached leaves were inoculated with 2 μl of a suspension containing 500 conidia of Botrytis cinerea strain B05.10. Conidia were preconditioned for 3 h in Gamborg’s B5 medium containing 10 mM sucrose and 10 mM potassium phosphate, pH 6. The adaxial surface of each leaf was inoculated at 10 different locations. Leaves were inserted into moist florist foam and incubated in clear plastic boxes under saturating humidity and low light conditions (Guimaraes et al. 2004). Frequency and size of expanding lesions were recorded 2 and 3 days post-inoculation. Non-expanding lesions can form a dry black necrosis, whereas expanding lesions are brown and water-soaked (Guimaraes et al. 2004). A caliper was used to measure lesion diameters.

Antifungal assay

In vitro assays were carried out as previously described (Li et al. 2001; Osborn et al. 1995). Conidia were harvested from sporulating cultures grown on potato dextrose agar (PDA) (Guimaraes et al. 2004). Conidia (5,000 ml−1) were suspended in potato dextrose broth. Soluble fractions of acetic ethanol extracts were generated from leaves as described above. Aliquots in 0.01% (v/v) acetic acid (50 μl) were incubated with conidia (50 μl) in wells of microtiter plates for less than 24 h at ~20°C. Fungal germination and growth were visualized with an Axiovert S 100 microscope (Carl Zeiss Inc., North America). Partial purification of foliar extracts was achieved using 3 and 30 kD Microcon molecular cut-off filters (Millipore, Billerica, MA). Magainin 2 (American Peptide Company, Sunnyvale, CA) was used as a positive control.

Microscopy

Pollen viability was assayed as previously described (Li et al. 2004). Anthers were removed from flowers at anthesis and squashed into a droplet containing 5 μg ml−1 fluorescein diacetate and 50 μg ml−1 propidium iodide. An Axiovert S 100 (Carl Zeiss Inc., North America) was used for fluorescence microscopy in combination with FITC (excitation: 495.5 nm, emission: 531 nm) and Texas Red (excitation: 571 nm, emission: 627 nm) filter sets. Callose was visualized after staining squashed anthers from immature flowers with 0.05% (w/v) aniline blue in 0.1 M sodium phosphate, pH 8.5. A DAPI (excitation: 402.5 nm, emission: 462 nm) filter set was used for microscopy.

Statistical tests

Averages and frequencies were analyzed with the statistics software SAS (Cary, NC) using ANOVA and a generalized linear model (GENMOD), respectively. Least square means or contrasts were used to statistically determine separation (α = 0.05). Homogeneity of variances was determined using Levene’s test.

Results

Endogenous expression of defensin mRNAs and peptides

DEF2 was abundantly expressed in immature floral buds at the meiotic stage (Goldberg et al. 1993) but rapidly declined thereafter (Fig. 1a). DEF2 was more abundant in petals and stamen than in sepals of 6–7 mm long floral buds. This pattern of expression was in stark contrast to DEF1 (Fig. 1b), clearly illustrating specific regulation of defensin gene expression during flower development. Conversely to DEF2, expression of DEF1 was low in stamen relative to petals and pistils, indicating that female and male reproductive organs preferentially express DEF1 and DEF2, respectively.

Fig. 1
figure 1

Defensin expression in developing flowers. a DEF2 mRNA levels in immature floral buds and organs of tomato cv. VF36, determined by RT-PCR. The gene was amplified for 30 cycles. Similar results were obtained after 20 or 25 cycles. Bud sizes define developmental stages. b DEF1 mRNA levels in immature floral buds and organs of tomato cv. VF36, determined by RT-PCR. Bud sizes indicate developmental stages. Plasmids containing DEF1 or DEF2 were used to demonstrate gene-specific amplification. The gene was amplified for 30 cycles. Similar results were observed after 20 or 25 cycles. c Actin mRNA levels in immature floral buds and organs of tomato cv. VF36, determined by RT-PCR. Amplification of genomic DNA shows that cDNA preparations were not contaminated. The gene was amplified for 30 cycles. d Expression and processing of soluble defensin peptides in immature floral buds and organs of tomato cv. VF36, determined by immunoblot analysis. Extracted proteins (2.5 μg) or recombinant defensins (0.5 μg) were separated. Recombinant DEF1 and DEF2 fusion proteins, cleaved with Factor Xa, were used as size standards for mature defensin peptides

Full size image

Mature peptides and pro-peptides of DEF1 and DEF2 are predicted to have molecular weights of 5.3 and 5.2 kD and 8.9 and 8.7 kD, respectively. As expected, mature forms of recombinant DEF1 and DEF2 peptides were recognized by an anti-defensin epitope antibody in form of ~5 kD bands (Fig. 1d). Lack of discrimination between DEF1 and DEF2 peptides by the antibody maybe explained by the fact that only 2 out of 10 amino acids differed between DEF1 and DEF2 epitopes. This antibody recognized peptides of >5 kD only in immature floral buds and stamen of tomato cv. VF36 (Fig. 1d). A ~5 kD band indicative of mature defensins accumulated at the meiotic stage in <3 mm long floral buds (Fig. 1d). Two antibody-reactive bands were observed in 3–5 mm long floral buds, most likely indicating expression of prodefensin, presumably containing the C-terminal domain, and mature defensin. The putative defensin precursor was also expressed in stamen from 6–9 mm long floral buds because only a band of approximately 9 kD was observed (Fig. 1d). The antibody used for detection did not discriminate between recombinant DEF1 and DEF2 peptides (Fig. 1d). The finding that antibody-reactive bands of >5 kD were not detected in carpels (Fig. 1d), is surprising because this organ is known to strongly express DEF1 mRNA (Milligan and Gasser 1995).

Effects of DEF2 mRNA expression on fertility of T0 and T1 generations

To investigate the function of DEF2, tomato cv. Zhongshu 5 was transformed separately with sense and antisense constructs of this gene. Ten transgenic plants were recovered that expressed the antisense transcript, but just three plants overexpressed the transgene. The transformation efficiency for the DEF2 sense construct was significantly lower, around 1% (data not shown), than the transformation efficiency for other genes, which is 30–40%. Endogenous DEF2 mRNA was detected only in two of the overexpressors (Fig. 2), suggesting cosuppression (Jorgensen 1995) in the remaining transformant. This orientation-dependent difference in recovery of transformants provided evidence that ectopic expression of defensin interfered with plant regeneration (χ2 = 5.33, P = 0.021).

Fig. 2
figure 2

DEF2 mRNA levels in leaves of individual transgenic tomato plants cv. Zhongshu 5 of the T0 generation as determined by RT-PCR. Sense plants are initialized S followed by numeric designators. Note that plant S20 does not express the transgene although it was confirmed as transformed by Southern blot hybridization. Plant S22 lacks detectable endogenous DEF2 expression perhaps due to cosuppression

Full size image

Transformant S5, one of the two plants ectopically expressing DEF2 (Fig. 2), was lost because it never set seed. Line S22 was not analyzed further because endogenous DEF2 mRNA expression was not detected (Fig. 2) and because no evidence of increased DEF2 protein expression was obtained by immunoblot analysis. Moreover, seed production and fruit weight was not significantly different from untransformed tomato (data not shown). The remaining line S8 and five antisense lines varying in DEF2 expression were selected for further analysis (Fig. 3a). The sense line S8 produced fewer seeds than normal, but the defect was less severe compared to transgenic plant S5 (Fig. 3b). Analysis of the T1 generation also yielded an inverse correlation between seed production and expression of antisense transcript (Fig. 3).

Fig. 3
figure 3

Reduced seed set and fruit size as a function of altered DEF2 expression. a Expression of DEF2 transgene in immature floral buds of tomato cv. Zhongshu 5 containing sense or antisense constructs as determined by RT-PCR. Sense and antisense lines of the T1 generation are initialized S and A, respectively, followed by numeric designators and are compared to untransformed (UT) plants. Actin was used as a loading control. Genes were amplified for 28 cycles. b Seed production of UT and transgenic tomato plants. Error bars indicate SEM (n ≥ 10). Asterisks indicate statistically significant differences from UT at P < 0.001 (***), P < 0.01 (**), or P < 0.05 (*). c Fruit weights of UT and transgenic plants. Error bars indicate SEM (n ≥ 11). Asterisks indicate statistically significant differences from UT at P < 0.001 (***) or P < 0.05 (*)

Full size image

Fruit maturation was delayed by more than 4 weeks in line S8 compared to untransformed tomato (t-test, P = 0.002). Fruits from line S8 were also significantly smaller than fruits from untransformed plants whereas the effect of antisense expression on fruit size was relatively minor (Fig. 3c).

Expression of DEF2 mRNA and defensin peptides during the T2 generation

Transgene expression was monitored by real-time PCR in the selected sense and antisense lines S8 and A2 (Fig. 4a). Constitutive expression of DEF2 driven by the cauliflower mosaic virus (CaMV) 35S promoter was observed during all floral stages analyzed. Constitutive expression of DEF2 mRNA in line S8 was less than tenfold higher than the endogenous expression of this gene in ~3 mm long untransformed floral buds (Fig. 4a). Simultaneously, endogenous DEF2 expression declined exponentially (Fig. 4a, b). No significant differences in endogenous DEF2 expression were observed in lines S8 and A2 (Fig. 4b).

Fig. 4
figure 4

DEF2 mRNA and DEF2 protein expression in transgenic tomato plants of the T2 generation. a Total DEF2 mRNA levels in untransformed (UT) and homozygous transgenic lines (S8 and A2), determined by real-time PCR. Total DEF2 mRNA levels in lines S8 and A2 primarily reflect transgene expression. Bud sizes define developmental stages. Results of an individual experiment are shown. b Endogenous DEF2 mRNA levels, determined by real-time PCR. Averages of three experimental replicates are shown for the earliest developmental stage; error bars indicate s.e.m. Averages of two experiments are shown for all other stages. c Expression of DEF2 protein in supernatant and pellet fractions of extracts from UT, S8, and A2 plants, determined by immunoblot analysis. Recombinant (abbreviated Rec.) DEF1 (0.2 μg) was used as a size marker. Note that processed DEF2 peptide of the expected size is constitutively expressed in sense line S8. No DEF2 peptide expression is observed in antisense line A2

Full size image

Mature DEF2 was expressed during all floral stages in line S8 (Fig. 4c). On the contrary, defensin was exclusively expressed in ~3 mm long floral buds of untransformed tomato cv. Zhongshu 5. Defensins, recovered from floral buds of untransformed plants and line S8, were insoluble at the meiotic stage. By comparison, defensin expression was absent from flowers of antisense line A2. DEF2 mRNA and protein levels were therefore not correlated in line A2.

Effect of altered DEF2 peptide expression on male reproductive development

Based on fertility defects observed during T0 and T1 generations (Fig. 3), we hypothesized that altered DEF2 expression interferes with male development. We therefore tested pollen viability in homozygous sense and antisense plants of the T2 generation (Fig. 5a and Supplemental Fig. S2). Pollen viability correlated with seed production in that line S8 produced a significantly smaller percentage of viable pollen grains than line A2 (Fig. 5b). No significant differences in meiosis and tetrad formation were observed in immature anthers from line S8 and untransformed plants. The only aberration we observed in meiotic products of line S8 was an infrequent production of supernumerary microspores (Fig. 5c). As microscopic examination of cross sections through 3–5 mm long floral buds from line S8 revealed little evidence of aberrations, the defect that was observed in mature pollen grains must have expressed itself after microsporogenesis. The exact time point when the pollen viability defect occurred was not determined because we did not analyze later stages of microgametogenesis. Nevertheless, we hypothesize that continuous expression of mature DEF2 in developing flowers (Fig. 4c) interferes with further development and maturation of pollen grains.

Fig. 5
figure 5

Defects in male reproductive development occurring in DEF2 transformants of the T2 generation. a Staining of alive and dead pollen grains from untransformed (UT) and homozygous transgenic (S8 and A2) tomato cv. Zhongshu 5 with fluorescein diacetate (green) and propidium iodide (red), respectively. The percentage of viable pollen at anthesis is shown below the images. b Pollen viability as a function of genotype. Error bars indicate s.e.m. (n = 7). Asterisks indicate statistically significant differences from UT at P < 0.001. c Aniline blue staining of callose in tetrads (UT and S8) and pollen mother cells (A2). Production of supernumerary microspores is indicated (arrow)

Full size image

The ratio of callose-containing pollen mother cells to tetrads was 6:1 for line A2 but 2:5 for untransformed tomato and line S8 when anther squashes from ~3 mm long floral buds were examined (Fig. 5c). Furthermore, two out of four flowers from line A2 were in meiosis or in the tetrad stage when cross-sections through developing anthers of ~3 mm long floral buds were examined, whereas none of four flowers from untransformed tomato were in meiosis but three flowers were in the tetrad stage. These observations indicate that silencing of defensin peptide expression (Fig. 4c) retards meiosis (Fisher’s exact test, P = 0.030).

Alteration of plant growth by ectopic expression of DEF2 peptide

Ectopic expression of DEF2 had pleiotropic effects on plant growth. Although growth was initially retarded, homozygous S8 seedlings caught up and eventually surpassed the growth of untransformed tomato plants (Fig. 6a). The stature of S8 plants differed from untransformed plants in that the leaves were smaller and growth was more upright, resulting in a more open architecture (Fig. 6b). Sepals of line S8 were shorter, making flowers appear stubby (Fig. 6c). The style length of line S8 was significantly shorter relative to untransformed plants (Fig. 6d, Supplemental Fig. S3A). No genotypic differences in the size and numbers of ovules were apparent (Supplemental Fig. S3B).

Fig. 6
figure 6

Pleiotropic effects of ectopic DEF2 expression on organ and plant growth. a Growth of untransformed (UT) and homozygous transgenic (S8 and A2) tomato cv. Zhongshu 5. Error bars indicate s.e.m. (n = 7). b Image of 8-week-old plants; bar, 10 cm. c Sepals of line S8 are shorter compared to UT tomato. d Image of carpels of flowers at anthesis; bar, 1 cm. e Style lengths as a function of genotype. UT, S8, and A2 tomato are compared. Error bars indicate s.e.m. (n ≥ 5). Asterisk indicates statistically significant differences from UT at P < 0.05

Full size image

Protection against B. cinerea by foliar expression of mature DEF2 peptide

Foliar expression of mature DEF2 (Fig. 7a) resulted in a significant decrease in disease incidence when tomato leaves were challenged with a conidial suspension of the pathogenic fungus Botrytis cinerea (Fig. 7b), consistent with the predicted antimicrobial activity of this peptide. Comparison of the relative antibody-reactive band intensities (Fig. 7a) suggests that DEF2 contributes to approximately 0.9% of the soluble protein in leaves of line S8, an expression level that is not dissimilar from the expression level of a radish defensin in transgenic tobacco (Terras et al. 1995). Despite this difference in disease incidence, the rate of lesion expansion was not altered. Noticeably, none of these changes in plant growth and defense affected line A2, illustrating a rather specific effect of antisense suppression on male reproductive development.

Fig. 7
figure 7

Constitutive DEF2 expression protects against the fungal pathogen B. cinerea. a Expression of soluble defensin peptides in leaves of untransformed (UT) and homozygous transgenic lines (S8 and A2), determined by immunoblot analysis. Extracted proteins (2.5 μg) or recombinant DEF1 (0.9 μg) were separated. The latter was used as a size marker. b Frequency of foliar infection 3 days post-inoculation by B. cinerea. Leaves from seven plants of each genotype were tested. Homozygous transgenic plants of the T2 generation (S8 and A2) were compared to untransformed (UT) plants (n = 7, means ± s.e.m.). Asterisks indicate statistically significant differences from UT at P < 0.001. The experiment was performed twice with similar results

Full size image

To further test the antifungal activity of DEF2, foliar extracts from transgenic lines or untransformed plants were incubated with conidia of B. cinerea. Soluble fractions inhibited hyphal growth (Fig. 8a). Surprisingly low protein concentrations were inhibitory. Foliar extract from the transgenic line S8 expressing DEF2 (Fig. 7a) was more active than those from the other two genotypes. Size fractionation was used to enrich for DEF2. Partial purification of molecules with a predicted size of 3–30 kD enhanced the genotype-dependent differences in antifungal activity (Fig. 8b). Whereas fractions from untransformed and antisense plants were not active, equivalent material from line S8 inhibited hyphal tip growth. It is likely that DEF2 is the active ingredient (i) because antifungal activity was only observed in size-fractionated material from line S8 and (ii) because DEF2 has a molecular mass of 5 kDa. Antifungal activity from line S8 differed from magainin, an antimicrobial peptide from the skin of Xenopus levis (Zasloff et al. 1988), in that the latter reduced the frequency of fungal germination, whereas the former disrupted fungal growth from the tip to the base. These effects were concentration-dependent. By comparison, fractions <3 kD did not contain antifungal activity (data not shown).

Fig. 8
figure 8

DEF2-overexpressing tomato leaves contain activity that inhibits growth of B. cinerea. a Soluble fractions of foliar extracts from untransformed (UT) or transgenic plants (S8 and A2) inhibit fungal growth; percentages of disrupted hyphal tips were quantified microscopically. b Micrographs of conidia incubated in 0.5 × potato dextrose broth without (0.005% acetic acid) or with increasing concentrations of magainin (12.5, 25, or 50 μg ml−1) or foliar extracts fractionated for molecular masses between 3 and 30 kD (0.75, 1.5, 3 OD280); bar, 25 μm. Disruption of hyphal tip growth is indicated (arrows)

Full size image

Discussion

Expression of defensin mRNAs and peptides

Our data suggest that different defensin genes are expressed in male or female organs (Fig. 1a, b). Surprisingly defensin was not found to be expressed in carpels of tomato cv. VF36 (Fig. 1d) although DEF1 mRNA is abundantly expressed in this floral organ (Milligan and Gasser 1995). The most likely explanation is that DEF1 expression is regulated at translational or post-translational levels because the antibody used reacted with recombinant peptides of both DEF1 and DEF2 (Fig. 1d).

Defensin peptide expression apparently differed between the tomato cultivars that were studied. Whereas defensins were detectable at different stages of tomato cv. VF36 development (Fig. 1d), these peptides seemed to be present only in <3 mm long floral buds of tomato cv. Zhongshu 5 (Fig. 4c). However, floral buds from the latter cultivar were not dissected, thus obviating enrichment of defensins in particular floral organs. Antigenic bands indicative of prodefensin were only in material from tomato cv. VF36, but apparent differences in regulation of the processing enzyme between cultivars during later stages of anther development are expected to be inconsequential assuming that the precursor peptide is inactive. Nevertheless, mature defensin peptides accumulated during the meiotic phase of flower development in both cultivars.

Insoluble rather than soluble defensins were recovered from floral buds at the meiotic stage (Fig. 4c), indicating a possible association of these peptides with plant cell wall components. The pectin-rich middle lamella of pollen mother cells thickens about tenfold at this stage (Gorman and McCormick 1997). PSORT, a program for predicting subcellular localization, assigned DEF2 to the cell wall (Nakai and Horton 1999). Nevertheless, different plant defensins are localized in the cell wall (Gao et al. 2000; Terras et al. 1995) or in the vacuole (Lay et al. 2003). Thus, we cannot exclude possible targeting of tomato defensins to the vacuole. Association with the pellet may be an extraction artifact caused by binding to insoluble aggregates, such as negatively charged pectic polysaccharides, which are likely to interact with cationic peptides, or starch grains.

Antisense suppression of defensin peptides (Fig. 4c) in homozygous A2 plants was not paralleled by changes in DEF2 mRNA expression (Fig. 4b) despite abundant expression of the transgene (Fig. 4a). Analysis of the primary transgenics also revealed that the majority of the antisense plants generated lacked detectable DEF2 peptide expression even though DEF2 mRNA was expressed (Supplemental Fig. 4). Significant positive correlation between sense and antisense expression was recently documented in plants using abiotic stress conditions and a whole-genome tiling array (Matsui et al. 2008a, b). It is possible that DEF2 antisense transcript blocks translation (Faghihi and Wahlestedt 2006). Alternatively, other types of regulation maybe involved as antisense RNAs participate in gene silencing, RNA stability, alternative splicing, RNA editing, RNA masking and methylation (Enerly et al. 2005; Hastings et al. 1997; Jen et al. 2005; Katiyar-Agarwal et al. 2006; Matsui et al. 2008a, b; Prescott and Proudfoot 2002; Tufarelli et al. 2003; Zubko and Meyer 2007).

Effects of altered DEF2 expression on male reproduction

Our results illustrate that DEF2 functions in male reproductive development. Phenotypic analysis of DEF2-silenced tomato plants implies that this defensin promotes meiosis in wild-type plants. Retardation of meiosis in DEF2-silenced tomato (Fig. 5c) apparently translates into a pollen viability defect similarly to late-acting sporogenous male sterile mutants (Gorman and McCormick 1997). Thus, DEF2 appears to function in developmental signaling. The only defensin-like peptide known to contribute to developmental signaling is the sterility locus cysteine-rich protein, which is expressed in pollen, but which has not been investigated for antimicrobial activity to our knowledge.

Inactivation of DEF2 during later stages of flower development is required for survival and development of pollen grains as continuous production of this peptide in transgenic tomatoes leads to a defect in male fertility. Alternatively, pleiotropic effects of constitutive DEF2 expression on plant growth (Fig. 6) may indirectly result in a pollen viability defect. However, the tapetum of plants ectopically expressing DEF2 appeared to be intact in anthers from 3–5 mm long floral buds (data not shown). We therefore favor the former possibility that ectopic expression of DEF2 has a direct detrimental effect on pollen grains during the later stages of anther development, including gametogenesis and pollen grain maturation. However, the CaMV 35S promoter used for constitutive expression of DEF2 is not active in male gametophytes after the first mitotic division (Van Der Leede-Plegt et al. 1992). The tapetum also degenerates at this stage (Gorman and McCormick 1997). Thus, the detrimental effect of DEF2 would have to express itself previously or be the result of persistently active peptide.

Antifungal properties of the DEF2 peptide

Two transformants were generated that ectopically expressed DEF2. One of them was lost because it never set seeds. The remaining line S8 produced relatively few seeds, but fertility was sufficient to generate homozygous offspring. Plants ectopically expressing DEF2 were recovered significantly less frequently than antisense transgenics. Constitutive DEF2 expression had pleiotropic effects on plant growth. Mature DEF2 expressed in leaves (i) reduced the frequency of B. cinerea infection and (ii) inhibited hyphal tip growth in vitro. Based on the DEF2 expression levels in leaves from line S8 (Fig. 7a) and concentrations of protein required to inhibit hyphal growth (Fig. 8), DEF2 appears to be as active as antifungal peptides from other plants (Osborn et al. 1995; Terras et al. 1995). Although the simplest explanation is that DEF2 itself is responsible for the observed antifungal activity, we cannot exclude the possibility that transgene expression induces other defence proteins or metabolites. Upregulation of other defence compounds could explain the relatively large effect of small DEF2 peptide levels (Fig. 7). Proteomics or transcriptomics would be a means to identify such induced defence pathways. Conclusive proof for the antifungal activity of DEF2 would require purification to homogeneity or analysis of a chemically synthesized and correctly folded peptide (Takayama et al. 2001). Nonetheless, our data clearly show that overexpression of DEF2 interferes with fungal invasion. Observations in line S8 may, however, not be entirely representative; recovery of two transgenics limited our ability to analyze constitutive DEF2 expression more broadly.

The mechanism of DEF2 action remains to be determined. However, certain plant defensins block L-type calcium channels (Spelbrink et al. 2004). Calcium dynamics have been studied during microsporogenesis and pollen development with highest levels of total Ca2+ during meiosis and the mature pollen stage (Tirlapur and Willemse 1992). Hyphal tip growth is dependent on a calcium gradient in the apical tip (Jackson and Heath 1993; Robson et al. 1991). While being speculative, our finding that partially purified extract from the transgenic line constitutively expressing DEF2 inhibits tip growth of B. cinerea is consistent with the idea that this defensin may alter Ca2+ transport. MsDef1 from Medicago sativa has been shown to disrupt Ca2+-dependent localization of Rab GTPase to root hair tips (Allan et al. 2008). Conversely, an epididymis-specific β-defensin induces sperm motility by stimulating Ca2+ uptake (Zhou et al. 2004).

Conclusion

Table 1 summarizes the effects of altered DEF2 expression on plant development and defence. Overexpression of DEF2 had pleiotropic effects on plant growth. Inhibition of early seedling growth (Fig. 6a) may reflect the low recovery rate of DEF2-overexpressing transformants. Although the role of DEF2 in development can only be tentative at this point because sense and antisense expression limited seed set (Fig. 3), these findings provide new insights into the regulation and functional diversification of antimicrobial plant defensins within the context of microsporogenesis, a developmental process that is still poorly understood (Yang et al. 2003).

Table 1 Summary of effects of DEF2 transgene expression on plant development and defense
Full size table