INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant defensins are a family of basic proteins of 45 to 54 amino acids that retard the growth of fungi, oomycetes, and gram-positive bacteria in vitro (Broekaert et al., 1995). The first plant defensins were isolated from wheat (Triticum aestivum) and barley (Hordeum vulgare), and were initially classified as a subgroup of the thionin family called the -thionins (Colilla et al., 1990; Mendez et al., 1990). Subsequent identification of other -thionin-like proteins in other plant families, together with structural information, revealed striking differences between -thionins and classical thionins (Bruix et al., 1993; Terras et al., 1995). Broekaert and colleagues (1995) later renamed this class of proteins as "plant defensins" due to structural and functional similarities with insect and mammalian defensins (Terras et al., 1995).

Plant defensins exhibit clear, although relatively limited, sequence conservation that is restricted to eight Cys that participate in four intramolecular disulfide bridges, two Gly at positions 13 and 34, an aromatic residue at position 11, and a Glu at position 29 (numbering relative to Rs-AFP2; Broekaert et al., 1995). Most plant defensins have been isolated from seeds (Broekaert et al., 1995), but they are also expressed in leaves (Terras et al., 1995; Segura et al., 1998), pods (Chiang and Hadwiger, 1991), tubers (Moreno et al., 1994; Stiekema et al., 1988), fruit (Meyer et al., 1996; Aluru et al., 1999), and floral tissues (Gu et al., 1992; Karunanandaa et al., 1994; Moreno et al., 1994; Milligan and Gasser, 1995; van den Heuvel et al., 2001; Park et al., 2002).

Many plant defensins have antifungal activity, although it is not known whether they have a common mode of action (Broekaert et al., 1997). One of the best-characterized antifungal plant defensins, Rs-AFP2 from radish (Raphanus sativus) seed, appears to act primarily at the cell membrane (Thevissen et al., 1996). Rs-AFP2 induces rapid Ca²⁺ uptake and K⁺ efflux from Neurospora crassa hyphae and thus may inhibit the growth of filamentous fungi by disrupting cytosolic Ca²⁺ gradients essential for hyphal tip growth (Thevissen et al., 1996). Thevissen and colleagues (1996, 2000) have suggested that the defensin initiates this response by interaction with a membrane-bound receptor rather than by permeabilizing the membrane by direct defensin-lipid interaction. Certain members within the plant defensin family also display other biological activities, including proteinase (Wijaya et al., 2000; Melo et al., 2002) and -amylase (Bloch and Richardson, 1991; Zhang et al., 1997) inhibitory activity and inhibition of protein translation (Colilla et al., 1990; Mendez et al., 1990; 1996) that may contribute to their role in defense.

Within the last decade, a number of defensin cDNA clones have been obtained using RNA from the floral tissues of solanaceous plants (Gu et al., 1992; Karunanandaa et al., 1994; Moreno et al., 1994; Milligan and Gasser, 1995; van den Heuvel et al., 2001). The proteins encoded by some of these clones are significantly different from the seed defensins because they have C-terminal domains of up to 33 amino acids in addition to the typical defensin domain of 47 to 48 amino acids. The encoded proteins have not been isolated previously, and their function has not been examined. Here, we describe the isolation and biological properties of a flower-derived defensin from the ornamental tobacco, Nicotiana alata (NaD1) as well as two related defensins from the petals of petunia (Petunia hybrida; PhD1 and PhD2).

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

cDNA Cloning of NaD1, PhD1, and PhD2

Three cDNAs encoding defensins were isolated from cDNA libraries prepared from floral tissues of ornamental tobacco and petunia. The cDNA from ornamental tobacco (NaD1) was obtained from a pistil cDNA library using a PCR product generated with oligonucleotide primers corresponding to a defensin cDNA clone from cultivated tobacco (Nicotiana tabacum; flower-specific thionin [FST]; Gu et al., 1992), whereas the two petunia cDNAs (PhD1 and PhD2) were isolated from a petal cDNA library during a screen for petal coloration genes. The cDNA clones encode predicted proteins of 105 amino acids for NaD1, and 103 and 101 amino acids for PhD1 and PhD2, respectively. All of the predicted proteins have a typical endoplasmic reticulum signal sequence of 25 amino acids followed by a defensin domain of 47 (NaD1 and PhD1) or 49 (PhD2) amino acids and a C-terminal domain of 27 to 33 amino acids.

An alignment of the deduced amino acid sequence of NaD1, PhD1, and PhD2 with that of other flower- and seed-derived defensins is shown in Figure 1A. NaD1, PhD1, and PhD2 and the floral defensins, FST from cultivated tobacco (Gu et al., 1992) and TPP3 from tomato (Milligan and Gasser, 1995) all have a predicted C-terminal domain of 27 to 33 amino acids in addition to the defensin domain. In contrast, the floral defensins PPT from petunia (Karunanandaa et al., 1994) and TGAS118 from tomato (van den Heuvel et al., 2001) lack this C-terminal domain, as do the seed defensins from radish (Rs-AFP2; [Terras et al., 1995], alfalfa [alfAFP; Gao et al., 2000]), and wheat (1-P; Colilla et al., 1990; Fig. 1A).

View larger version (58K): [in this window] [in a new window]   Figure 1.   A, Alignment of the predicted amino acid sequence of NaD1 (accession no. AF509566), PhD1 (accession no. AF507975), and PhD2 (accession no. AF507976) with the predicted amino acid sequences encoded by four other flower-derived cDNA clones: FST from tobacco (accession no. Z11748), TPP3 from tomato (Lycopersicon esculentum; accession no. U20591), PPT from petunia (accession no. L27173), TGAS118 from tomato (accession no. AJ133601), and the purified seed defensins Rs-AFP2 from radish (accession no. P30230), alfAFP from alfalfa (Medicago sativa; accession no. AF31946), and 1-P from wheat (accession no. P2015). GenBank accession numbers are given in parenthesis. The endoplasmic reticulum signal sequence has been omitted. Identical residues are boxed in black with conservative substitutions in gray. Spaces have been introduced to maximize the alignment. The arrow indicates the site of cleavage between the mature defensin and C-terminal prodomain. The disulfide bond connectivities are shown below the sequences as connecting solid lines. The additional disulfide bond in PhD1 and PhD2 is shown by a broken line. B, Comparison of the basic (Arg, Lys, and His) and acidic (Glu and Asp) amino acid composition and the net charge associated with the defensin and C-terminal domains in NaD1, PhD1, PhD2, FST, and TPP3 at neutral pH.

The central defensin domain of NaD1 (Fig. 1A, amino acids 1-47) shares 60% and 69% amino acid identity with PhD1 and PhD2, respectively, whereas PhD1 and PhD2 share 76% amino acid identity. This contrasts to the sequence identity obtained for pair-wise comparisons of the defensin domain of NaD1 with the other solanaceous defensins, FST (98%) and TPP3 (63%), the "tail-less" floral defensins, PPT (38%) and TGAS118 (36%), and the seed defensins, Rs-AFP2 (31%), alfAFP (31%), and 1-P (32%). Overall, only about 23% of the amino acids in the defensin domain are conserved between these plant defensins. These are restricted to the eight Cys (positions 4, 15, 21, 25, 36, 45, 47, and 51), two Gly at positions 13 and 34, an aromatic residue at position 11, and a Glu at position 29 (numbering relative to Rs-AFP2). PhD1 is an exception because the conserved Gly-13 and Glu-29 residues are replaced with a Ser and Ala, respectively. Furthermore, PhD1 and PhD2 have an extra pair of Cys residues at positions 8 and 24 (numbering relative to Rs-AFP2; Fig. 1A).

Analysis of the charge distribution within the defensin and C-terminal domains of NaD1, PhD1, PhD2, FST, and TPP3 revealed that the defensin domain carries a net positive charge, whereas the C-terminal domain is dominated by acidic and hydrophobic amino acids and carries a net negative charge at neutral pH (Fig. 1B). It is interesting that the positive charge associated with the defensin domain is essentially neutralized by the acidic C-terminal domain.

Temporal and Spatial Expression of NaD1 in Ornamental Tobacco

Temporal and spatial expression of the NaD1 gene was initially examined by RNA-blot analysis. The NaD1 transcript (0.6 kb) was detected in anthers, pistils (stigma and style), ovaries, and petals from ornamental tobacco flowers, but was barely detectable in roots and was not detected in leaves (data not shown). In addition, transcript levels were substantially higher in stage I flowers compared with mature flowers. Thus, in situ hybridization experiments were performed to examine NaD1 expression in stage I flower buds. The transcript was most abundant in the epidermal cell layers of the petals and sepals, within the connective cells of the anthers, and the cortical cells of the style. No transcript was detected in the tapetum, pollen mother cells, the transmitting tissue, or the vascular bundles of the anther and style (Fig. 2).

View larger version (107K): [in this window] [in a new window]   Figure 2.   In situ location of NaD1 mRNA in flower buds. A and B, Autoradiographs of transverse sections of a stage I (10 mm in length) ornamental tobacco flower from after hybridization with a 35S-labeled NaD1 antisense RNA probe. The epidermal (ep) cells of the petal (pe) and sepal (se), and the cortical cells (cc) of the style (st) and the connective tissue (ct) of the anther (a) were heavily labeled. There was no hybridization to the pollen mother cells (pmc), tapetum (ta), vascular bundle (vb), or the transmitting tissue (tt). C, Autoradiograph of a transverse section of a stage I flower bud after hybridization with a 35S-labeled NaD1 sense RNA probe. The cells of the style (st), anther (a), petal (pe), and sepal (se) were not labeled.

The NaD1 Precursor Is Processed to Release the Mature Defensin Domain

Polyclonal antibodies were raised in a rabbit to bacterially expressed NaD1 proprotein (6H.proNaD1; central defensin domain plus C-terminal domain). The antibodies bound specifically to two proteins of approximately 5 and 7 kD in buffer-soluble protein extracts derived from whole ornamental tobacco flowers and various floral tissues that had been collected at different stages of development (Fig. 3, A and B). In all cases, the level of the approximately 7-kD protein declined relative to the level of the approximately 5-kD protein with flower maturation (Fig. 3, A and B). Moreover, these immunoreactive proteins appeared to be more abundant in extracts from floral tissues collected at the earlier stages of development than at maturity (Fig. 3, A and B).

View larger version (70K): [in this window] [in a new window]   Figure 3.   Immunoblot analysis of NaD1 in ornamental tobacco flowers and floral tissues at various stages of development. A, Five stages (I-V) of flower development as described in "Materials and Methods." Immunoblot of buffer soluble proteins (30 µg lane1) derived from petals, anthers, pistils, ovaries, and sepals at stages I through V using the -6H.proNaD1 antibodies. B, Immunoblot of buffer soluble proteins (30 µg lane1) derived from whole flowers at stages I through V, 500 and 1,000 ng of purified NaD1 (M), and 5 and 10 ng of purified 6H.proNaD1 (P) using the -6H.proNaD1 antibodies. Molecular mass markers are in kilodaltons. The positions of NaD1 (M) and proNaD1 (P) are marked in with arrows.

To test the relative affinity of the -6H.proNaD1 antibodies to NaD1 and 6H.proNaD1, various amounts of both proteins were immunoblotted (Fig. 3B), and the intensities of the immunoreactive protein bands were quantified by densitometric analysis. The antibody reaction was approximately 140-fold stronger for 6H.proNaD1 compared with NaD1 (Fig. 3B). Densitometric analysis was also applied to an immunoblot of buffer-soluble extracts derived from whole flowers at stages I through V of development to determine the relative amount of mature NaD1 present in these extracts (Fig. 3B). NaD1 represented approximately 4% (w/w) of the soluble protein extracted from stage I flower buds and approximately 1% (w/w) in stage V flowers. This translates to estimates of approximately 0.36 and 0.09 mg of NaD1 gram¹ of fresh weight of floral tissues at stage I and V, respectively.

Purification of Mature Floral Defensins from Ornamental Tobacco and Petunia

The mature NaD1 protein was extracted from flowers (stages I-IV) in 50 mM sulfuric acid and purified using ammonium sulfate precipitation, heat treatment, and gel filtration. Fractions containing NaD1 were identified by SDS-PAGE and immunoblot analysis with the -6H.proNaD1 antibodies. Proteins in the immunoreactive fractions were resolved further by reverse phase (RP)-HPLC (Fig. 4A). Peak 1 was identified as the mature (approximately 5 kD) NaD1 protein by immunoblot analysis (Fig. 4A, inset), mass spectrometry, and N-terminal amino acid sequencing (Fig. 4C). The mass of 5,296.7 D compared favorably with the mass of 5,296.0 D for the mature defensin domain predicted from the cDNA clone when all four disulfide bonds have formed. The protein peaks that eluted ahead of peak 1 were identified as the 6-kD proteinase inhibitors (NaPIs) that have been previously described by Atkinson et al. (1993) and were confirmed by immunoblot analysis using -NaPI-specific antibodies (data not shown). No other defensin-related proteins were identified in the other RP-HPLC peaks.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Purification of defensins from ornamental tobacco flower buds and petunia petals. RP-HPLC profile of gel filtration fractions from ornamental tobacco (A) and petunia (B) extracts showing percentage buffer B (%B) and retention times in minutes. Defensin peaks collected are numbered. The inset in A is an immunoblot of protein in peak 1 after incubation with -6H.proNaD1 antibodies. C, N-Terminal sequence and electrospray mass spectrometry data for the proteins in peaks 1 (A), 2 and 3 (B), and the predicted mass of the defensin domains encoded by the NaD1, PhD1, and PhD2 cDNA clones. "x" corresponds to an unassigned amino acid that is probably a Cys as predicted from the cDNA sequence.

Two defensins were isolated from petunia petals using the same purification procedure except that the gel filtration fractions that contained the defensins were identified by SDS-PAGE and Coomassie Blue staining rather than by immunoblot analysis. Gel filtration fractions containing approximately 5-kD proteins were pooled and subjected to RP-HPLC. Two major peaks (Fig. 4B) that eluted at 28.2% and 29.5% (v/v) acetonitrile had the same mass and N-terminal sequence as the defensins encoded by the PhD1 and PhD2 cDNA clones, respectively (Figs. 1A and 4C). The masses also revealed that all 10 Cys residues were oxidized.

Intracellular Location of NaD1

Given that the NaD1 gene encoded a preproprotein with a signal sequence and a C-terminal prodomain, we investigated whether the mature protein was secreted like the seed defensins that lack this domain or was directed to an intracellular compartment such as the vacuole. The -6H.proNaD1 antibodies were used in immunogold electron microscopy on 10-µm sections of anthers and ovaries collected from stage I flower buds. NaD1 was located within electron dense aggregates within the vacuole and was not detected in the cytoplasm or the cell walls (Fig. 5).

View larger version (87K): [in this window] [in a new window]   Figure 5.   Immunogold localization of NaD1 in anthers and ovaries from stage I flowers. A, Overview of the anther showing cells of the connective tissue with electron dense deposits (arrowed) in the vacuole (v). B, Connective tissue cells of the anther and the cortical cells (C) of the ovary labeled with the -6H.proNaD1 antibodies. The antibodies bound specifically to electron dense deposits in the vacuole and no binding was observed in the cytoplasm (cy) or cell walls (cw).

Antifungal Activity of NaD1

Defensins purified from the flower buds of ornamental tobacco and petunia petals were tested in vitro for antifungal activity against Fusarium oxysporum and Botrytis cinerea. The defensins inhibited the growth of both fungi, however, they were more effective against F. oxysporum (Fig. 6A) than B. cinerea (Fig. 6B). At 2 µg mL¹, PhD1 and NaD1 inhibited the growth of F. oxysporum by 56% and 42%, respectively, whereas PhD2 had no effect. At 10 µg mL¹, PhD1 and NaD1 completely inhibited growth, whereas inhibition by PhD2 was 86%. In contrast, all three defensins were ineffective at 2 µg mL¹ against B. cinerea. NaD1 was the best inhibitor at 10 µg mL¹ (96%), followed by PhD1 (70%) and PhD2 (41%). Ovalbumin and the 6-kD proteinase inhibitors (NaPI) from ornamental tobacco had no inhibitory effect at the concentrations tested, the latter showing that fungal inhibition is not a feature of all small, Cys-rich proteins.

View larger version (35K): [in this window] [in a new window]   Figure 6.   Effect of NaD1, PhD1, and PhD2 on the growth of F. oxysporum f. sp. dianthi Race 2 (A) and B. cinerea (B). Growth of the fungi in the test solutions is plotted relative to the growth in water for the same period (50 h). Test proteins were used at final concentrations of 2, 10, and 20 µg mL1. Growth in water is taken to be 100% growth. Ovalbumin and the 6-kD proteinase inhibitors from ornamental tobacco (NaPI) were used as negative controls. Each treatment was performed in quadruplicate. The error bars are SE of the mean.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

NaD1 is a member of a distinct class of defensins that are produced by several members of the Solanaceae, including ornamental tobacco, cultivated tobacco, petunia, and tomato (Fig. 1). These defensins differ from most plant defensins because they are produced from precursor proteins with a C-terminal prodomain of 27 to 33 amino acids. The best-characterized members of this class of defensins are highly expressed in developing flower buds (Gu et al., 1992; Milligan and Gasser, 1995), however, they are also induced in salt-stressed leaves of Nicotiana species (Komori et al., 1997; Yamada et al., 1997) and in the fruits of Capsicum chinense (Aluru et al., 1999). Similar prodomains have been described for the thionins (Bohlmann, 1994) and mammalian (Michaelson et al., 1992; Yount et al., 1995) and insect defensins (Lowenberger et al., 1999).

The role of the C-terminal prodomain in the floral defensins is not known. However, a number of potential roles have been described for the prodomains on thionins as well as on mammalian defensins. These include roles in subcellular (i.e. vacuolar) targeting, as intramolecular steric chaperones for folding, and/or in preventing detrimental interactions between the mature protein domain and other proteins or lipid membranes during intracellular trafficking (Michaelson et al., 1992; Bohlmann, 1994; Florack and Stiekema, 1994; Florack et al., 1994). The latter two roles were suggested based on the observation that the cationicity of the mature protein domain could effectively be neutralized by the anionic prodomain at neutral pH (Michaelson et al., 1992). The charge distribution of the defensin and prodomains in the solanaceous defensins is consistent with this observation (Fig. 1B).

We used polyclonal antibodies raised against bacterially expressed prodefensin from ornamental tobacco to confirm that whole flowers and dissected floral tissues produce an immunoreactive protein of the size expected for proNaD1 (approximately 7 kD). Furthermore, the level of proNaD1 in these tissues decreased relative to the approximately 5-kD mature NaD1 with flower maturation, suggesting processing of proNaD1 to NaD1. When various amounts of purified NaD1 and 6H.proNaD1 were immunoblotted with the -6H.proNaD1 antibodies, a much stronger antibody reaction (approximately 140-fold) was observed with 6H.proNaD1 (the immunizing antigen). Thus, this result suggests that the relative intensities of the approximately 7- and 5-kD protein species are not a reflection of the relative amounts of proNaD1 and mature NaD1, but higher affinity of the antibodies for the unprocessed form (proNaD1).

NaD1 and the petunia defensins (PhD1 and PhD2) are stable under extremes of pH and temperature. These properties were exploited in the purification process that involved extraction in 50 mM sulfuric acid, heating at 90°C for 10 min, and exposure to organic solvents during the RP-HPLC purification. Their inherent stability arises from the characteristic defensin structure known as the Cys-stabilized (CS) motif (Broekaert et al., 1997; Fant et al., 1999; Lay et al., 2003). This motif consists of an -helix that is packed against a triple-stranded antiparallel -sheet and is stabilized by four disulfide bonds. The petunia defensins are unusual because they have an additional disulfide bond. Therefore, we propose that PhD1 and PhD2 define a new subclass of plant defensins.

We have solved the solution structures of NaD1 (Lay et al., 2003), PhD1, and PhD2 (B.J.C. Janssen, H.J. Schirra, F.T. Lay, M.A. Anderson, D.J. Craik, unpublished data) by ¹H NMR spectroscopy. The additional disulfide bond in PhD1 and PhD2 provides a further stabilizing covalent bond between the -helix and the loop between the -helix and the first -strand. The floral defensins have the same three-dimensional structure as the seed defensins even though they share little sequence identity and are produced from precursor proteins with C-terminal prodomains.

The floral defensin in ornamental tobacco is located in the vacuole unlike the seed defensins such as Rs-AFP2 from radish seeds, which is located in the middle lamellae (Terras et al., 1995) and alfAFP from alfalfa, which is also extracellular (Gao et al., 2000). This indicates that a role in vacuolar targeting should be considered for the C-terminal prodomain on NaD1, although this needs to be confirmed experimentally because there are no consensus sequences that define C-terminal vacuolar-sorting determinants (Neuhaus and Rogers, 1998). However, the C-terminal prodomain of NaD1 is rich in acidic and hydrophobic amino acids, a common feature of C-terminal vacuolar-sorting determinants (Nielson et al., 1996). The C-terminal prodomains of PhD1, PhD2, FST, and TPP3 share similar properties to the NaD1 prodomain and may also direct these proteins to the vacuole. Given that the C-terminal "tail" is not present in most plant defensin precursors, roles in protein folding and/or detoxification as proposed for the prodomains of mammalian defensins and thionins seem unlikely, but cannot be disregarded.

It should also be noted that given the higher affinity of the antibodies to proNaD1 compared with NaD1, other putative locations for mature NaD1 aside from the vacuole cannot be excluded, as these populations may not be detected by the antibodies under the conditions used in the experiment.

In the thionins, processing of the precursor protein involves at least two steps: the cotranslational cleavage of the endoplasmic reticulum signal peptide and the posttranslational removal of the C-terminal prodomain (Ponz et al., 1983; Romero et al., 1997). Romero et al. (1997) previously demonstrated that the prodomain of the barley leaf DG3 thionin is processed in the vacuole by a 70-kD proteinase and that the mature DG3 thionin accumulates in the vacuolar content. This observation reinforces the idea that the prodomain may serve as a vacuolar targeting determinant for correct deposition and processing of the thionin precursor in the vacuole. Thus, by analogy, similar events could be involved in the trafficking and processing of the NaD1 precursor.

During the initial library screens for floral defensin cDNA clones, only one clone was obtained from ornamental tobacco and two from petunia. This was unexpected because defensins are often members of multigene families. For example, there are up to 15 different defensin genes in Arabidopsis (Thomma et al., 2002) and several FST-related genes have been identified in cultivated tobacco (Gu et al., 1992). It is interesting that the proteins that were purified in this study corresponded exactly to the isolated cDNA clones and were abundant in ornamental tobacco and petunia flowers. No other defensins were detected in the protein extracts, indicating that if related defensin genes are present, they are expressed at much lower levels.

Expression of NaD1 is similar to that described previously for FST (Gu et al., 1992) and TPP3 (Milligan and Gasser, 1995). In all cases, expression is highest in young floral buds and decreases substantially as the flower matures. NaD1, FST, and TPP3 are all expressed in petals, anthers, and pistils, but are not detectable in leaves. The NaD1 transcript accumulates in the epidermal cells of the sepals and petals and, interestingly, in the cortical cells of the style and the connective cells of the anther. It is not present in the tapetum, pollen mother cells, or the transmitting tissue of the style that guide and nurture pollen tubes on their way to the ovary. It is interesting that whereas FST is also expressed in the cortical cells of the style, expression in the anther is restricted to the anther wall. In addition, it is not expressed in sepals, and expression in petals is restricted to the adaxial surface of the petals rather than both sides as occurs with NaD1.

More recently, van den Heuvel and coworkers (2001) used in situ hybridization to examine the expression of TGAS118, a floral defensin from tomato that lacks the C-terminal prodomain. In contrast to NaD1, this "tail-less" defensin is expressed in tissues that are in contact with, or develop into germ cells. Like NaD1, TGAS118 is produced by the epidermal cells of the petal, however, within the anther, TGAS118 transcript accumulates in the tapetum and pollen mother cells, as well as the connective and middle layer cells. In the pistil, TGAS118 transcript is present in the transmitting tissue and cortical cells (van den Heuvel et al., 2001).

Expression of NaD1 in the outermost layers of the sepals and petals and in pistil and anther tissues that surround, but do not come into direct contact with pollen or pollen tubes as they grow through the style, is consistent with a role in protecting the valuable germ cells against damage by potential pathogens. Furthermore, this pattern of expression suggests that NaD1 is unlikely to have a role in pollen-pistil interactions as suggested for the defensin-like molecule PCP-A1 from cauliflower (Brassica oleracea; Doughty et al., 1998). The potential role in defense was supported by the observation that NaD1 and the two floral defensins from petunia inhibited the growth of two fungal pathogens in vitro. The defensins were more active against F. oxysporum than B. cinerea in the in vitro assays. Growth inhibition of 42% to 56% at 2 µg mL¹ against F. oxysporum and 41% to 96% at 10 µg mL¹ against B. cinerea compares favorably with the IC₅₀ values of 2 to 11 µg mL¹ for the radish defensins on F. culmorum and B. cinerea (Terras et al., 1995).

Although it is difficult to directly determine the concentration of the defensins in vivo because they are concentrated in specific cell layers in the flowers, we partially addressed this issue by estimating the levels of mature NaD1 in buffer-soluble extracts from whole flowers. A conservative estimate of about 0.36 mg of NaD1 g¹ fresh weight of floral tissue or 4% (w/v) of total soluble protein at the earliest bud stage was made. Thus, the protein is abundant and the concentrations in the expressing cell layers are probably higher than those that were effective in the in vitro assays.

It was interesting to note that the additional disulfide bond in PhD1 and PhD2 did not result in significant differences in the antifungal activity of these proteins compared with NaD1. However, it is likely that it confers additional thermostability and rigidity to the proteins.

Flowers of solanaceous plants appear to produce at least two classes of defensins. One class is produced as a preproprotein with a C-terminal prodomain that is deposited in the vacuole, and the second class is produced without the C-terminal prodomain and is found extracellularly. Expression of the defensins in different floral tissues and potential differences in subcellular location indicate that defensins may serve a role in protection of reproductive tissues against a range of potential pathogens.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Material

Ornamental tobacco (Nicotiana alata) plants of mixed self-incompatibility genotype were maintained under greenhouse conditions. Organs were collected from flowers and floral buds within 2 h of harvest. Whole flowers were harvested at developmental stages defined as follows: stage I: closed bud (5-10 mm in length), stage II: elongated bud (20-30 mm in length), stage III: elongated bud with emerging petals (40-50 mm in length), stage IV: elongated bud with emerging pigmented petals (60-70 mm in length), and stage V, fully open pigmented flower (70-80 mm in length). Petunia (Petunia hybrida) petals were collected from petunia var. Old Glory Blue (Ball Seed, Chicago) that had been grown and maintained under greenhouse conditions. All samples were stored at 80°C until use.

Cloning of cDNAs Encoding Floral Defensins from Ornamental Tobacco and Petunia

RNA was extracted from ornamental tobacco pistils (stage III/IV) using Trizol reagent and the protocol from Invitrogen (Carlsbad, CA). Single-stranded cDNA was prepared using the Superscript Preamplification System (Invitrogen) and was amplified by PCR using oligonucleotide primers FST1a (5'-GGAATTCCATATGGCTCGCTCCTTGTGC-3') and FST1b (5'-GCGGATCCTCAGTTATCCATTATCTCTTC-3') that correspond to the DNA sequence published for the FST (Gu et al., 1992) from cultivated tobacco (Nicotiana tabacum). Primers FST1a and FST1b matched the sequence of FST between nucleotides 49 and 66 and 346 and 363 and incorporated EcoRI and BamHI restriction sites, respectively. The PCR was performed for 30 cycles with the following temperature profile: 95°C, 30 s; 55°C, 1 min; 72°C, 1 min, with a final cycle modified to include an extension time of 5 min at 72°C. The amplified product (336 bp) was isolated from agarose gels and was cloned into pBluescript II SK⁺ vector (Stratagene, La Jolla, CA) for sequencing. The clone, designated pBS-NaD1, was subsequently used to screen an ornamental tobacco (S₆S₆) pistil cDNA library (Schultz et al., 1997). The petunia defensin clones, PhD1 and PhD2, were obtained from a petal cDNA library during a screen for petal coloration genes (Brugliera et al., 1994).

In Situ Hybridization

In situ hybridization was performed on 10-mm ornamental tobacco flower buds essentially as described by Li et al. (1999). ³⁵S-Labeled sense and antisense RNA probes were produced by linearizing the pBS-NaD1 DNA with EcoRI and BamHI and transcribing with T7 and T3 RNA polymerases (Promega, Madison, WI), respectively.

Bacterial Expression of the Ornamental
Tobacco Defensin

A DNA fragment encoding the NaD1 proprotein (proNaD1, precursor minus the N-terminal endoplasmic reticulum signal sequence) was amplified by PCR using the pBS-NaD1 plasmid as template and oligonucleotide primers NaT1 (5'-CCGGATCCAGAGAATGCAAAACAG-3') and NaT3 (5'-GGGAGCTCTTAGTTATCCATTATCTC-3') that incorporated a BamHI and SacI restriction site, respectively. Conditions for the PCR were as described for the cDNA cloning. The amplified PCR product was initially cloned into the pGEM-T (Promega) vector and was then subcloned into the BamHI and SacI sites of the pQE30 (Qiagen, Valencia, CA) vector for protein expression with an N-terminal hexa-His tag in Escherichia coli strain M15 (Qiagen). The expressed protein (6H.proNaD1) was purified from lysed bacterial cells using the denaturing protein purification protocol outlined in the TALON Metal Affinity Resin User Manual (PT1320-1; CLONTECH Laboratories, Palo Alto, CA). Bound protein was eluted from the resin with 100 mM EDTA, pH 8.0, lyophilized and purified further by RP-HPLC on a Brownlee RP300 C8 column (4.6 × 100 mm; Perkin-Elmer, Norwalk, CT) using a pump (model 510; Waters, Milford, MA) and detector (model 481UV; Waters). Samples were applied in 0.1% (v/v) trifluoroacetic acid in water (buffer A) and were eluted with 60% (v/v) acetonitrile in 0.089% (v/v) trifluoroacetic acid (buffer B) using a linear gradient of 0% to 100% (w/v) buffer B over 30 min at a flow rate of 1 mL min¹. The identity of the purified protein was confirmed by mass spectrometry and N-terminal sequencing.

Production of a Polyclonal Antiserum

Purified 6H.proNaD1 (1.3 mg) was conjugated to keyhole limpet hemocyanin (0.3 mg; Sigma, St. Louis) with glutaraldehyde as described by Harlow and Lane (1988) and injected (100 µg) into a rabbit with an equal volume of Freund's complete adjuvant (Sigma). Booster immunizations were administered 5 and 9 weeks later, and consisted of protein conjugate (100 µg) mixed with Freund's incomplete adjuvant (Sigma). Preimmune serum was collected before injection and immune serum was collected 9 d after the second immunization. The immunoglobulin (Ig) G fraction in the preimmune serum and immune serum was purified on Protein-A Sepharose CL-4B (Amersham Pharmacia Biotech, Piscataway, NJ) and was stored at a final concentration of 6.4 mg mL¹ at 80°C.

SDS-PAGE and Immunoblot Analysis

Buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, and 0.5 M NaCl)-soluble protein extracts were prepared from whole flowers and dissected floral tissues from stages I through V of development (3 mL of buffer g¹ fresh weight). The protein concentration was determined using a protein assay kit (Bio-Rad, Hercules, CA) with bovine serum albumin as standard. Protein samples (30 µg lane¹) were subjected to SDS-PAGE on 15% (w/v) polyacrylamide gels (Laemmli, 1970) and immunoblot analysis using -6H.proNaD1 antibodies (1:2,500 dilution). Bound antibody was detected using a donkey -rabbit IgG conjugated to horseradish peroxidase (1:3,500 dilution; Amersham Pharmacia Biotech) and enhanced chemiluminescence detection reagents (Amersham Pharmacia Biotech) before being exposed to Hyperfilm (Amersham Pharmacia Biotech).

Isolation of Floral Defensins

Defensins were extracted from flowers using a modification of the procedure for extraction of thionins from barley (Hordeum vulgare) flour (Ozaki et al., 1980). Whole ornamental tobacco flowers up to the petal coloration stage of flower development (5-70 mm, 650 g wet weight) were ground to a fine powder with liquid nitrogen using a mortar and pestle and were processed further in an Ultra-Turrax homogenizer (Janke and Kunkel, Staufen, Germany) in 50 mM sulfuric acid (3 mL g¹ wet weight). After stirring for 1 h at 4°C, insoluble material was removed by filtration through Miracloth (Calbiochem, Alexandria, Australia), followed by centrifugation (25,000g, 15 min, 4°C). The slurry was adjusted to pH 7.8 by the slow addition of 10 M NaOH and was stirred for 1 h at 4°C before removal of precipitated material by centrifugation (25,000g, 15 min, 4°C). Solid ammonium sulfate was added to 80% (w/v) saturation and the mixture was stirred for 4 to 16 h at 4°C to precipitate the defensin protein. The precipitate was collected by centrifugation and was dissolved in 50 mL of gel filtration buffer (150 mM KCl and 10 mM Tris-HCl, pH 8.0) before heating at 90°C for 10 min. After centrifugation, the supernatant was loaded onto a Sephadex G-50 (medium grade; Amersham Pharmacia Biotech) gel filtration column (85 × 2.54 cm). Fractions (50 mL) were collected and analyzed by immunoblotting with -6H.proNaD1 antibodies. Fractions containing NaD1 were pooled, concentrated by rotary evaporation at 45°C, and filtered through a 0.22-µm syringe filter (Millipore, Bedford, MA) before further purification by RP-HPLC.

RP-HPLC was performed on a System Gold HPLC (Beckman, Fullerton, CA) coupled to a 166 detector (Beckman). Analytical RP-HPLC was conducted on a Brownlee Aquapore RP300 C8 column (4.6 × 100 mm; Perkin-Elmer), whereas preparative runs were performed using a C8 column (22 × 250 mm; Vydac, Hesperia, CA). The protein was eluted with a linear gradient of 0% to 100% (v/v) buffer B (60% [v/v] acetonitrile in 0.089% [v/v] trifluoroacetic acid) at a flow rate of 1 or 10 mL min¹ over 40 min, respectively.

The same procedure was used for the purification of defensins from the petals of petunia except that after gel filtration and SDS-PAGE, the gels were stained with Coomassie Blue and fractions containing proteins of approximately 5 kD were pooled and subjected to RP-HPLC. Defensin peaks were identified by mass spectrometry and N-terminal sequencing. The defensins were concentrated by rotary evaporation and lyophilization and were redissolved in water at 1 mg mL¹ and stored at 20°C until use.

Electrospray Ionization Mass Spectrometry and
N-Terminal Amino Acid Sequencing

Electrospray ionization mass spectrometry was conducted with 1 to 100 pmol of RP-HPLC-purified protein in 2 to 4 µL of 50% (v/v) acetonitrile containing 0.1% (v/v) formic acid using a Sciex API-300 triple quadruple (Perkin-Elmer) fitted with a microion spray ion source and analyzed using Sciex Biomultiview software (Perkin-Elmer). N-Terminal amino acid sequence was obtained by sequential Edman degradation using a automated protein sequencing system (G1005A; Hewlett-Packard, Palo Alto, CA).

Densitometric Analysis of Immunoblots

Various amounts of purified NaD1 and 6H.proNaD1 were subjected to SDS-PAGE on a Novex precast 4% to 12% (w/v) Bis-Tris polyacrylamide gel (Invitrogen) and were immunoblotted using the -6H.proNaD1 antibodies (1:2,500 dilution). The exposed immunoblot film was scanned using a flatbed scanner (Scanjet 5p; Hewlett-Packard) and the image was imported into the ImageQuaNT software (version 4.2a; Molecular Dynamics, Sunnyvale, CA) for densitometric analysis.

Fixation and Immunogold Labeling for
Electron Microscopy

Anthers and ovaries were removed from stage I ornamental tobacco flower buds and were fixed in 4% (w/v) formaldehyde and 0.5% (w/v) glutaraldehyde in 60 mM PIPES/KOH, pH 7.2, for 2 h at room temperature and then overnight at 4°C. After fixation, the tissues were washed in 60 mM PIPES/KOH, pH 7.2, and dehydrated for 3 h at room temperature in acidified dimethoxypropane (concentrated hydrochloric acid:dimethoxypropane, 1:2,000 [v/v]). The dehydrated segments were embedded in LR Gold containing Benzil (London Resin Company, Berkshire, UK) by polymerization under a UV lamp (TUV 15-W; Phillips, Mahwah, NJ) at a distance of 10 cm for 12 h at 25°C.

Immunogold labeling of ultrathin sections was performed as described in Anderson et al. (1987). The protein A-purified -6H.proNaD1 antibodies were incubated with sections at a final concentration of 64 or 21 µg of IgG mL¹ for anther and ovary sections, respectively. Specificity of labeling was tested by replacing the primary antibody with antibodies purified from preimmune serum at the same concentration. For visualization of ultrastructure, the sections were stained for 15 min in 3% (w/v) aqueous uranyl acetate and for 2 min with Sato triple lead stain (Sato, 1968) before being viewed on a electron microscope (Phillips CM120; FE1 Company, Eindhoven, The Netherlands).

Fungal Growth Inhibition Assays

The 96-well microtiter plate assay of Broekaert et al. (1990) was used to test the effect of purified NaD1, PhD1, and PhD2 on the growth of Botrytis cinerea (isolated from rose petals by Florigene, Melbourne, Australia) and Fusarium oxysporum (f. sp. dianthi, Race 2; isolated from carnation by Florigene). Fungal spores were isolated from sporulating cultures growing on one-half-strength potato dextrose agar (Difco Laboratories, Detroit, MI). The spore suspension was filtered through two layers of sterile muslin, and the concentration was determined using a hemocytometer. The spore concentration was adjusted to 2 × 10⁴ spores mL¹ in potato dextrose broth (Difco Laboratories). Spores were used directly or after storage in sterile 20% (v/v) glycerol solution at 20°C.

The spore suspension (80 µL) was added to the wells of a sterile 96-well flat-bottomed microtiter plate (Greiner, Frickenhausen, Germany) to which 20 µL of filter-sterilized (0.22-µm syringe filter; Millipore) test protein (10, 50, or 100 µg mL¹) or water was added. Ovalbumin was from Sigma and the 6-kD Cys-rich proteinase inhibitors (NaPI) were purified from ornamental tobacco stigmas as described by Atkinson et al. (1993). The purity and concentration of each protein was confirmed before use by RP-HPLC analysis. The plates were shaken on an orbital shaker to mix the spores and test solution. The plates were incubated at 22°C in darkness, and the optical density of each well was determined using a microplate reader (Spectra Max Pro 250; Molecular Devices) set at 595 nm over a period of 50 h. Each sample was performed in quadruplicate.