Mutational analysis of a plant defensin from radish (Raphanus sativus L.) reveals two adjacent sites important for antifungal activity.

Mutational analysis of Rs-AFP2, a radish antifungal peptide belonging to a family of peptides referred to as plant defensins, was performed using polymerase chain reaction-based site-directed mutagenesis and yeast as a system for heterologous expression. The strategy followed to select candidate amino acid residues for substitution was based on sequence comparison of Rs-AFP2 with other plant defensins exhibiting differential antifungal properties. Several mutations giving rise to peptide variants with reduced antifungal activity against Fusarium culmorum were identified. In parallel, an attempt was made to construct variants with enhanced antifungal activity by substituting single amino acids by arginine. Two arginine substitution variants were found to be more active than wild-type Rs-AFP2 in media with high ionic strength. Our data suggest that Rs-AFP2 possesses two adjacent sites that appear to be important for antifungal activity, namely the region around the type VI β-turn connecting β-strands 2 and 3, on the one hand, and the region formed by residues on the loop connecting β-strand 1 and the α-helix and contiguous residues on the α-helix and β-strand 3, on the other hand. When added to F. culmorum in a high ionic strength medium, Rs-AFP2 stimulated Ca2+ uptake by up to 20-fold. An arginine substitution variant with enhanced antifungal activity caused increased Ca2+ uptake by up to 50-fold, whereas a variant that was virtually devoid of antifungal activity did not stimulate Ca2+ uptake.

During the last decades, it has been recognized that many living organisms produce small antimicrobial peptides to protect their tissues from infectious microbial agents. Well known examples of peptides with antimicrobial properties are the cecropins of invertebrates (reviewed in Ref. 1) and magainins of amphibians (reviewed in Ref. 2). Another class comprises cysteine-rich peptides, among which are the mammalian and insect defensins (3)(4)(5), both small, basic proteins with a cysteine-stabilized three-dimensional folding pattern involving antiparallel ␤-sheets. Insect defensins are produced upon perception of pathogens by the insect fat body and are secreted in the hemolymph (5). Mammalian defensins are present in phagocytic blood cells and are also produced by epithelial cells of the intestines and airways (4). Antimicrobial peptides have also been found in plants. Thionins, for instance, are highly basic 5-kDa peptides toxic to both Gram-positive and Gram-negative bacteria, fungi, yeast, and various mammalian cell types (6). A number of plant species produce thionins constitutively in their seed as well as in their leaves in a pathogen-inducible way (7). Other potent antimicrobial peptides found in plants are structurally related to defensins of mammals and insects and are therefore termed plant defensins (8). Plant defensins are small cysteine-rich peptides consisting of 45-54 amino acids with four intramolecular disulfide bridges. They are encountered in different plant species and various tissues such as seed, flowers, and pathogen-stressed leaves. Comparison of the known primary sequences of a series of plant defensins shows that the arrangement of the cysteines is highly conserved and reveals the existence of a cysteine-stabilized ␣-helix motif (9), which is also present in insect defensin A (10). Their three-dimensional structure consists of three antiparallel ␤-strands and an ␣-helix (11) and is similar to that of insect defensins (5) and some scorpion toxins (e.g. charybdotoxin; Ref. 12). Most plant defensins hitherto isolated exhibit antifungal activity. Some of them, for example SI␣2 1 (Sorghum bicolor inhibitor 2 of ␣-amylase), are inhibitors of ␣-amylases but do not inhibit fungal growth (13,14). The plant defensins with antifungal activity can be divided in two groups. The first group causes morphological distortions of the fungal hyphae resulting in swollen and hyperbranched fungal structures (9,14). The second group merely inhibits fungal growth without inducing morphological changes. Mode of action studies performed on a representative of each class (Rs-AFP2 from radish and Dm-AMP1 from dahlia) has shown that plant defensins cause rapid ion fluxes upon addition to fungal hyphae, resulting in Ca 2ϩ uptake, K ϩ efflux, and medium alkalinization (15).
In this study we have performed a structure-function analysis of Rs-AFP2, a plant defensin isolated from radish seed and member of the plant defensin group causing hyperbranching of fungal hyphae. It is the most potent among a number of plant defensin isoforms occurring in radish, including Rs-AFP1 isolated from seed and Rs-AFP3 and Rs-AFP4 isolated from in-fected leaves (8). In order to investigate which amino acids are essential for the antifungal activity of Rs-AFP2, we have undertaken a mutational analysis of this peptide. Information on amino acid substitutions resulting in either a decreased or enhanced antifungal activity, taken together with preliminary data on the three-dimensional configuration of Rs-AFP2, allows prediction of the sites which possibly interact with the yet unknown target site on the fungal hyphae.
Site-directed Mutagenesis-Mutagenesis of the Rs-AFP2 coding sequence was performed by two sequential polymerase chain reactions (PCR) as described in Ref. 18. Primers used in the PCR reaction are listed in Table I. In a first PCR, part of the Rs-AFP2 coding sequence was amplified using a sense mismatch primer containing the desired mutation and primer OWB35, a derivative of the M13 reverse primer elongated with a 5Ј tag (28 cycles; 1 min at 94°C, 1 min at 55°C, 1 min at 72°C). For design of the mismatch primer, the yeast preferential codon usage was taken into account (19). Ten ng of PvuI-linearized plasmid pBluescript/RsAFP* (20) was used as a template for the first PCR. The amplified product containing the mismatch served as a megaprimer to further elongate the Rs-AFP2 sequence (5 cycles; 1 min at 94°C, 1 min at 55°C, 1 min at 72°C). In a second PCR, this elongated fragment was amplified by primer OWB61, binding to the 5Ј end of the Rs-AFP2 gene, and OWB36, an oligonucleotide identical to the 5Ј tag of OWB35 (28 cycles; 1 min at 94°C, 1 min at 55°C, 1 min at 72°C). OWB61 contains a restriction site allowing in-frame cloning into the HindIII site in the MF␣1 pro-sequence region of pVD4 (20). Amplification products of the second PCR were digested with HindIII-BamHI and introduced in the corresponding sites of pVD4. After verification of the occurrence of the desired mutations by nucleotide sequence determination, the expression blocks containing the MF␣1 promoter and preprosequence followed by the mutated Rs-AFP2 gene were isolated by SalI-BamHI restriction digestion and subcloned into the SalI-BglII-digested yeast shuttle vector pTG3828 (21). After subcloning, the sequence of the mutated Rs-AFP2 domain was verified by nucleotide sequencing. Re-striction enzymes were purchased from Boehringer Mannheim (Mannheim, Germany), T4 DNA ligase from Life Technologies, Inc. (Life Technologies, Merelbeke, Belgium), and Taq DNA polymerase from Appligene (Pleasanton, CA). DNA sequencing was performed on a Pharmacia A.L.F. DNA sequencer using the AutoRead Sequencing Kit (Pharmacia, Uppsala, Sweden) according to the manufacturer's instructions.
Heterologous Expression and Purification of Rs-AFP2 Variants-Transformation of S. cerevisiae, growth of the yeast cultures, and purification of the Rs-AFP2 variants from the culture supernatants were essentially done as described previously for native Rs-AFP2 (20). Briefly, 250 ml of culture supernatant (minimal selective SD medium: 0.8 g/liter CSM-URA from BIO 101, La Jolla, CA; 6.5 g/liter yeast nitrogen base from Difco; 20 g/liter glucose (Merck); 5 g/liter casamino acids from Difco) was passed over an anion-exchange chromatography column (Q-Sepharose Fast Flow, Pharmacia) connected on-line with a disposable reversed phase C 8 silica column (Bond Elut, 500 mg solid phase, Varian, Harbor City, CA). The C 8 silica column was subsequently rinsed with 6 ml of 10% (v/v) acetonitrile containing 0.1% (v/v) trifluoroacetic acid. The Rs-AFP2 variants were eluted from the latter column with 4 ml of 30% (v/v) acetonitrile containing 0.1% (v/v) trifluoroacetic acid. After drying in a rotating vacuum concentrator, the eluted fractions were purified by reversed-phase chromatography on a C 2 /C 18 silica column (Pep-S, 5-m beads, 0.4 ϫ 25 cm, Pharmacia). Fractions were collected manually, and the elution position of the Rs-AFP2 variants was determined by a combination of antifungal activity analysis on F. culmorum in SMFϪ (synthetic medium fungi; Ref. 17) and SDS-PAGE analysis. In all cases, elution positions could be determined unambiguously.
Protein Analysis-Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was carried out according to Ref. 22 using a 15% (w/v) acrylamide, 0.5% (w/v) bisacrylamide separating gel and a 5% (w/v) acrylamide, 0.1% (w/v) bisacrylamide stacking gel. Gels were either stained with Coomassie Brilliant Blue R250 or immunoblotted using anti-Rs-AFP1 antibodies as described previously (8). Protein concentrations were determined by the bicinchoninic acid method (23) using authentic Rs-AFP2 as a standard. Free cysteine thiol groups were determined by the Ellman assay on both reduced and unreduced protein samples as described previously (17). Circular dichroism spectra were obtained on a Jasco 600 spectropolarimeter with a cell path of 0.02 cm. Proteins were dissolved at 0.5 mg/ml in distilled water. The spectra were acquired in a single scan mode (10 nm/min) in the ultraviolet region of 265-185 nm. Circular dichroism data were base line-corrected and are presented in units of ⌬⑀ (M Ϫ1 cm Ϫ1 ) (24).
Large Scale Purification of Rs-AFP2 Variants-Recombinant yeast (S. cerevisiae) cells containing vectors for expression of Rs-AFP2(Y38G) and Rs-AFP2(V39R), respectively, were grown for 7 days in a fermentor 5Ј-ATCTGCTACCGTCCTTGTTAATAG-3Ј (Biostat E 15 liter) at 25°C as a batch-fed culture and harvested at a final OD of approximately 80. Cells were pelleted by centrifugation at 3000 ϫ g for 20 min. The supernatants were passed directly over a Biopilot S-Sepharose column (15 ϫ 10 cm, Pharmacia) pre-equilibrated in 20 mM ammonium acetate, pH 6. A flow rate of approximately 100 ml/min was maintained using gravity feed. The bound fraction was eluted with a single wash of 1 liter of 500 mM ammonium acetate (pH 6) and freeze-dried for 3 days to completely remove the ammonium acetate salt. The freeze-dried fraction was dissolved in 20 mM ammonium acetate and refractionated by cation exchange chromatography on a S-Sepharose Fast Flow column (10 ϫ 2.6 cm, Pharmacia) equilibrated in 20 mM ammonium acetate, pH 6. The bound fraction was eluted with a linear gradient of 20 -500 mM ammonium acetate (pH 6) over 325 min at 3 ml/min. Proteins were monitored by on-line measurement of absorbance at 280 nm. Fractions containing Rs-AFP2 variants were identified either by using a standard in vitro antifungal bioassay (see below) or by Western blot (see above). Fractions containing the expressed peptide were pooled, freeze-dried, and further purified by reversed phase high performance liquid chromatography on a Pep-S column (C 2 /C 18 silica, 25 ϫ 0.93 cm, Pharmacia). Peptides were eluted with a linear gradient of 0.1% (v/v) trifluoroacetic acid to 99.9% (v/v) acetonitrile, 0.1% (v/v) trifluoroacetic acid over 100 min at a flow rate of 3 ml/min. A single predominant peak of absorbance at 280 nm containing the Rs-AFP2 variants was eluted at approximately 20% (v/v) acetonitrile.
Antifungal Activity Assays-The antifungal activity assays were carried out in microplates as described in Ref. 25. A 2-fold dilution series of the protein in sterile water was prepared, and 20 l of the serial dilutions were added to 80 l of synthetic low ionic strength medium (SMFϪ, Ref. 17) containing 10 4 spores/ml of the test fungus without or with the addition of extra salts as indicated under "Results." The plates were incubated at room temperature. Growth of the fungi was monitored microscopically after 24 h and by microspectrophotometry after 72 h unless otherwise indicated. The protein concentration required for 50% growth inhibition (IC 50 value) was calculated as described in Ref. 17. The specific antifungal activity was defined as 1/IC 50 . Preliminary Three-dimensional Solution Structure Determination by 1 H NMR-All 51 residues of Rs-AFP1 have been sequence specifically assigned following the strategy of Wü thrich (26) using a combination of double quantum filtered correlation spectroscopy (DQF-COSY) (27), nuclear Overhauser effect spectroscopy (NOESY) (28), and homonuclear Hartmann-Hahn spectroscopy (HOHAHA) (29) spectra recorded on a Bruker AM-500 (Bruker Analytische Messtechnik, Karlsruhe, Germany). These were recorded at a protein concentration of 1.3 mM and pH 4.2, in both 9/1 H 2 O/D 2 O and D 2 O solutions at two different temperatures (304.5 and 313.2 K). The data set at 312.2 K was used to extract 775 nuclear Overhauser effect cross-peaks, and 44 3 J NH␣ and 62 3 J ␣␤ coupling constants. Using the programs CALIBA, HABAS, and GLOMSA (30), 775 upper limit constraints, 94 angle constraints, and 19 stereospecific assignments were generated. These data, together with 13 upper and 13 lower limits for the disulfide bridges and the pyroglutamate ring, were used to calculate 500 structures with the program DIANA using the REDAC protocol (30) on a SG Crimson (Crimson, Mountain View, CA). The 25 best structures, with a root mean square deviation of the backbone atoms of all the 51 residues of 1.15 Ϯ 0.22 Å and a root mean square deviation of all the heavy atoms of 1.80 Ϯ 0.24 Å, were optimized by simulated annealing (31, 32) using DISCOVER (AMBER forcefield) and INSIGHT II for visualization (Biosym Technologies, Inc., San Diego, CA). The root mean square deviation of these final structures is 1.33 Ϯ 0.26 Å for the backbone atoms and 1.84 Ϯ 0.28 Å for all the heavy atoms.

Conception and Production of Rs-AFP2 Variants-Proteins
belonging to the family of plant defensins have been purified and sequenced from a range of taxonomically divergent plant species, while others have been identified via cDNA sequencing (9). Fig. 1 represents a comparison of the complete amino acid sequences of 12 different plant defensins whose antifungal activity against F. culmorum has been assessed in our laboratories (8,14,17,33). In terms of biological activity, three groups of plant defensins can be discerned: group I, those who are inhibitory to F. culmorum and cause increased hyphal branching; group II, those inhibitory to F. culmorum without causing hyphal deformations; and group III, those not affecting growth of F. culmorum at concentrations below 100 g/ml. Rs-AFP2, the protein studied in this work, belongs to the first group.
As can be seen from the alignment of the sequences in Fig. 1, the pattern of cysteines is totally conserved in all the sequences, as is the glycine residue at position 34 (numbering relative to the studied protein Rs-AFP2). Those residues are important secondary structure elements and are part of the cysteine-stabilized ␣␤ motif characterized by the sequences CXXXC, GXC, and CXC (X stands for any amino acid) (34). Other well conserved residues are the serine at position 8, the glycine at position 13, and the glutamate at position 29. Those conserved residues were not considered for substitution in the present study, since it is likely that they play a role in determining the structure of the peptide.
A number of amino acid residues were found to be fully conserved among the antifungal plant defensins (group I and II) but subject to non-conservative changes in plant defensins devoid of antifungal activity (group III). Those residues, namely Gln-5, Thr-10, Gly-16, and Ala-31, were considered to be suitable candidate residues for site specific mutational analysis. Lys-44 and Tyr-48, which are conserved in all group I and II plant defensins, except Dm-AMP1 and Hs-AFP1, respectively, were also retained for mutational analysis. In addition, amino acids that are conserved in group I but not in group II could be important for causing the typical morphological deformation of fungal hyphae, which is characteristic for group I plant defensins. These residues comprise Tyr-38, Phe-40, and Pro-41 and were likewise selected for mutational analysis.
A first series of Rs-AFP2 variants was conceived such that the amino acid residues selected as discussed above were substituted by the corresponding residue of SI␣2, a group III plant defensin devoid of antifungal activity. Residue Pro-41 was deleted rather than substituted as the loop between the ␤-strand 2 and ␤-strand 3 comprising Pro-41 is shorter in group III plant defensins than in group I plant defensins and, furthermore, contains no proline residue in SI␣2. It was expected that some of these substitution variants would have a reduced antifungal activity versus wild-type Rs-AFP2.
A second series of Rs-AFP2 variants was conceived in such a way that amino acids at selected positions were replaced by the basic amino acid residue arginine. The underlying rationale for these substitutions is that Rs-AFP1, the near-identical but less basic natural analogue of Rs-AFP2, has a lower antifungal activity than Rs-AFP2, especially when assessed in media with a high ionic strength (17). Rs-AFP1 and Rs-AFP2 only differ at two residues (Gln-5 is Glu and Arg-27 is Asn in Rs-AFP1), both of which substitutions result in a higher net positive charge of Rs-AFP2 versus Rs-AFP1. This suggests that an increase in the net charge of Rs-AFP2 by replacement of certain residues with arginine might further increase its antifungal activity. The positions selected for the arginine substitutions were those that show weak conservation among the different plant defensins or that are occupied by basic residues in plant defensins other than Rs-AFP2 (see Fig. 1).
For the production of the different Rs-AFP2 variants with the desired amino acid substitution, the Rs-AFP2 coding sequence was mutated site-specifically by PCR, fused in frame to the yeast mating factor ␣1 (MF␣1) promoter and prepro-sequence (20,21) and subsequently transferred to yeast via a yeast shuttle vector. The different Rs-AFP2 analogues were purified from the yeast culture supernatant by a combination of ion-exchange chromatography and reversed-phase chromatography. Using this approach, we have previously shown that wild-type Rs-AFP2 can be produced in a correctly processed and bioactive form in yeast (20). In total, 19 Rs-AFP2 variants were produced and purified in this way (see Table II). The purity of the preparations was assessed by SDS-PAGE analysis. All Rs-AFP2 variants migrated essentially as single bands which had the same electrophoretic mobility as wild-type Rs-AFP2 (Fig. 2). In addition, all purified proteins were recognized by anti-Rs-AFP1 antiserum on immunoblots prepared from SDS-PAGE gels, confirming their identity as variants of Rs-AFP2 (results not shown).
Two Rs-AFP2 variants, Rs-AFP2(Y38G) and Rs-AFP2(V39R) with a substitution of the tyrosine at position 38 by glycine and of valine at position 39 by arginine, respectively, were purified on a large scale from 15-liter fermentation cultures of the appropriate recombinant yeast strains. Circular dichroism spectroscopic studies were performed on these variants as well as on authentic Rs-AFP2. The circular dichroism spectrum of  Rs-AFP2(V39R) was virtually identical to that of Rs-AFP2, indicating that neither the substitution itself nor the way the variant was synthesized in yeast had imposed alterations of backbone secondary structure elements (Fig. 3). Rs-AFP2-(Y38G) had a circular dichroism spectrum which was almost identical to that of Rs-AFP2, except for a slightly decreased steepness of the ⌬⑀ drop in the 190 -208 nm region (Fig. 3). Antifungal Activity of Rs-AFP2 Variants-The purified Rs-AFP2 substitution variants were assessed for their antifungal activity against F. culmorum in two different media: a low ionic strength medium called SMFϪ (17), and the same medium supplemented with 1 mM CaCl 2 and 50 mM KCl, called SMFϩ. The presence of salts in the test medium, especially salts with divalent cations, is known to reduce the specific antifungal activity of Rs-AFP2 (17). Seed-purified as well as yeast-expressed wild-type Rs-AFP2 served as controls in the assays. The results of these comparative tests, expressed as IC 50 values, are presented in Table II. Most of the variants of the first series (substitutions by corresponding SI␣2 residues) showed no or only a minor decrease of their antifungal activity in medium SMFϪ, except Rs-AFP2(A31W), Rs-AFP2(Y38G), and Rs-AFP2(P41⌬), which showed a substantial decrease in antifungal potency. In SMFϩ, the medium with added salts, a significant decrease in antifungal activity was observed for the following Rs-AFP2 analogues of the first series: Rs-AFP2-(T10G), Rs-AFP2(A31W), Rs-AFP2(Y38G), Rs-AFP2(F40M), Rs-AFP2(P41⌬), and Rs-AFP2(K44Q). On the other hand, the substitutions Q5M and G16M resulted in a slight but significant increase in antifungal potency, especially noticeable in medium SMFϩ, whereas the substitution Y48I had little or no effect on the antifungal activity.
In contrast to what was expected, most of the Rs-AFP2 variants of the second series (arginine substitutions) did not show an enhanced antifungal activity compared to Rs-AFP2. In some cases, an even lower antifungal activity was observed, possibly caused by the unfavorable presence of a positive charge at that position or by the absence of a residue necessary for interaction with the fungal target. The largest decrease in antifungal activity was observed for substitution variants Rs-AFP2(L28R) and Rs-AFP2(I46R), whereas variants Rs-AFP2(S12R), Rs-AFP2(I42R), and Rs-AFP2(I49R) showed only a modest reduction in antifungal activity. However, two Rs-AFP2 variants, namely Rs-AFP2(G9R) and Rs-AFP2(V39R), were about 2-fold more active than wild-type Rs-AFP2 when assessed in SMFϩ.
The antifungal activity of Rs-AFP2(V39R) purified from a large scale culture of recombinant yeast was further characterized in SMF with increasing Ca 2ϩ or K ϩ concentrations and compared with that of authentic Rs-AFP2 (isolated from seed) as well as yeast-purified Rs-AFP2. As is shown in Fig. 4, the antifungal activity against F. culmorum of Rs-AFP2(V39R) was less reduced by the presence of cations in the growth medium in comparison with wild-type Rs-AFP2. Indeed, in the presence of 5 mM CaCl 2 and at a concentration of 10 g/ml, Rs-AFP2(V39R) caused complete inhibition of the growth of F. culmorum, whereas wild-type Rs-AFP2 was basically inactive under the same conditions. At 10 g/ml, wild-type Rs-AFP2 was fully active against F. culmorum only when the CaCl 2 concentration was equal or lower than 1.25 mM. Likewise, the activity of wild-type Rs-AFP2 was drastically reduced in the presence of 100 mM KCl, whereas Rs-AFP2(V39R) was still fully inhibitory to F. culmorum at this KCl concentration.
The potency of Rs-AFP2(V39R) relative to authentic Rs-AFP2 was also assessed on a set of seven different phytopatogenic fungi in three media differing in ionic strength: SMFϪ, SMF including 1 mM CaCl 2 and 50 mM KCl (SMFϩ), and SMF including 5 mM CaCl 2 and 50 mM KCl. As can be seen from the data presented in Table III, the relative antifungal activity of the variant was dependent on the test organism. On three fungi (F. culmorum, N. hematococca, and V. dahliae), Rs-AFP2(V39R) was more active than Rs-AFP2. In the medium SMFϩ, for  instance, Rs-AFP2(V39R) was about 2-, 5-, and 5-fold more potent than Rs-AFP2 against F. culmorum, N. hematococca and V. dahliae, respectively. As in this medium neither Rs-AFP2 nor Rs-AFP2(V39R) inhibited growth of A. brassicicola, A. pisi, or B. cinerea at concentrations below 50 g/ml, the highest concentration tested, no difference in antifungal potency could be observed for these fungi. However, on P. betae, Rs-AFP2(V39R) was less potent than Rs-AFP2. The differences in antifungal potency between Rs-AFP2(V39R) and Rs-AFP2 were always more pronounced in the SMF media with added salts than in the low ionic strength medium SMFϪ. Effect of Rs-AFP2 Variants on Ca 2ϩ Uptake by Fungi-Although the precise molecular target of Rs-AFP2 on fungal hyphae is not yet known, recent work in our laboratory has shown that Rs-AFP2 causes very rapid ion fluxes, including increased Ca 2ϩ uptake, when added to fungal hyphae (15). To investigate whether the ability of Rs-AFP2 to stimulate Ca 2ϩ uptake in fungi is linked to its antifungal effect, 45 Ca 2ϩ uptake was measured in F. culmorum treated with different concentrations of either Rs-AFP2, the virtually inactive variant Rs-AFP2(Y38G), and the variant with increased antifungal potency, Rs-AFP2(V39R). The medium used for this test consisted of half-strength potato dextrose broth supplemented with 1 mM MgCl 2 and 50 mM KCl. As shown in Fig. 5, Rs-AFP2 caused a dose-dependent increase of 45 Ca 2ϩ uptake, which at a dose of 100 g/ml reached a level that was about 20-fold higher relative to water-treated controls. At the same dose, Rs-AFP2-(V39R) stimulated 45 Ca 2ϩ uptake by over 50-fold, and the higher 45 Ca 2ϩ uptake stimulation of Rs-AFP2(V39R) versus wild-type Rs-AFP2 was observed over the whole concentration range tested. In marked contrast, however, addition of the variant Rs-AFP2(Y38G) with impaired antifungal properties resulted in 45 Ca 2ϩ uptake rates that fluctuated around the levels observed for water-treated control cultures. DISCUSSION A structure-activity analysis of Rs-AFP2, a plant defensin from radish causing growth inhibition of fungal hyphae (17), was carried out in order to investigate which residues are important for antifungal activity of the peptide. Candidate amino acid residues were considered to be those conserved among plant defensins exhibiting antifungal activity but not among those devoid of antifungal activity as outlined in Fig. 1. Following this rationale, we have chosen to produce a series of nine Rs-AFP2 analogues in which particular amino acid residues were changed to the corresponding residue of the plant defensin SI␣2, which does not display antifungal activity. Residue Pro-41 was deleted rather than substituted as the loop encompassing this residue is shorter in SI␣2 than in Rs-AFP2. A second series of Rs-AFP2 variants was aimed at increasing the net positive charge (at physiological pH) of Rs-AFP2 by substituting particular residues by an arginine at various nonconserved positions along the Rs-AFP2 sequence. This approach was inspired by the fact that Rs-AFP2, which has a higher net positive charge than Rs-AFP1, has a 2-30-fold higher activity relative to Rs-AFP1 (17).
Wild-type and variant peptides were produced in yeast and purified using identical chromatographic procedures. After the last purification step consisting of reversed phase chromatography, the different peaks were assayed for antifungal activity in order to identify the elution position of the Rs-AFP2 variant. All peptides showed similar retention times. When analyzed by SDS-PAGE, all variant peptides showed the same electrophoretic mobility as wild-type Rs-AFP2, indicating that they have approximately the same size. The structural conformation was studied into more detail by circular dichroism spectroscopy for the variants Rs-AFP2(V39R) and Rs-AFP2(Y38G). In the case of Rs-AFP2(V39R), the circular dichroism spectrum was virtually identical to that of Rs-AFP2, whereas the spectrum of Rs-AFP2(Y38G) showed a slightly altered spectrum in the 190 -280 nm region. This alteration is most probably due to the conformational flexibility of glycine within a polypeptide chain. The presence of glycine in the type VI ␤-turn connecting ␤-strand 2 and ␤-strand 3 may entail some relaxation of this region and loosen the packing of the ␤-sheet. The conformation of the other variants was not verified, but the absence of free thiol groups indicated that the disulfide bridges had formed.
Within the first substitution series, variants that showed a clearly reduced activity on F. culmorum were Rs-AFP2(T10G), Rs-AFP2(A31W), Rs-AFP2(Y38G), Rs-AFP2(F40M), and Rs-AFP2(P41⌬). The importance of the residues at positions 10, 38, and 40 is underscored by our observation that substitution variants in which those residues were replaced by an alanine showed a similar drop in antifungal activity. 2 In the second series, consisting of arginine substitution variants, additional variants were identified that displayed reduced antifungal activity, namely Rs-AFP2(S12R), Rs-AFP2(L28R), Rs-AFP2-(A42R), and Rs-AFP2(I46R). It is not clear whether the reduction in antifungal activity of these variants was due to the unfavorable presence of an extra charge or to the replacement of an amino acid essential for the antifungal activity.
Remarkably, the loss in antifungal potency in all these cases was less noticeable in the low ionic strength medium than in the medium supplemented with 1 mM CaCl 2 and 50 mM KCl. This may be explained by assuming that the interaction between Rs-AFP2 and its putative receptor on fungal hyphae is based both on ionic interactions and non-ionic stereospecific interactions. Upon increasing the ionic strength of the medium, the ionic interactions with the putative receptor are weakened due to competition between Rs-AFP2 and inorganic cations. In the case where non-ionic stereospecific interactions are weakened due to an unfavorable substitution, the overall interaction is also expected to become more susceptible to ionic competition.
The two most interesting Rs-AFP2 analogues of the arginine substitution series are Rs-AFP2(G9R) and Rs-AFP2(V39R). Al-though these variants show no significantly increased activity on F. culmorum in the low ionic strength medium, their activity on this fungus is much less influenced by the presence of cations in comparison with wild-type Rs-AFP2. This is again consistent with our model, which predicts that the interaction between Rs-AFP2 and its putative receptor is based both on ionic and non-ionic interactions. Introducing an extra charged residue at positions 9 or 39 may reinforce the ionic interactions, leading to variants that are at an advantage in competing with cations for binding at the putative receptor site.
The relative antifungal potency of the arginine substitution variant Rs-AFP2(V39R) compared to Rs-AFP2 appeared to be dependent on the test fungus. Rs-AFP2(V39R) was more active on F. culmorum, N. hematococca, and V. dahliae (three taxonomically related fungi belonging to the family Nectriaceae), but less active on P. betae. This suggests that the putative receptor on hyphae of different fungal species may reveal conformational or compositional differences.
As relatively high ionic strength conditions occur in all plant cell compartments (17), Rs-AFP2 variants such as Rs-AFP2(G9R) and Rs-AFP2(V39R) displaying a decreased cation antagonism in their activity against some phytopatogenic fungi could be useful for plant transformation experiments aimed at obtaining disease-resistant crops. We have previously shown that transgenic tobacco plants expressing wild-type Rs-AFP2 are more resistant to the fungal pathogen Alternaria longipes than untransformed plants (8). Further enhancement of the resistance level may be achieved through the expression of either Rs-AFP2(G9R) or Rs-AFP2(V39R) in transgenic plants.
We have previously shown that Rs-AFP2 stimulates Ca 2ϩ uptake by fungal hyphae, an effect that can be observed within minutes after addition of the peptide (15). This stimulation of Ca 2ϩ uptake may be part of the responses triggered by the interaction of Rs-AFP2 with its putative receptor. Our results now seem to indicate that antifungal activity and ability to trigger enhanced Ca 2ϩ uptake are correlated. Indeed, the variant Rs-AFP2(Y38G), which is virtually devoid of antifungal activity in presence of inorganic salts, was unable to stimulate Ca 2ϩ uptake in F. culmorum. On the other hand, the arginine substitution variant Rs-AFP2(V39R) displaying enhanced antifungal potency caused about 2.5-fold higher Ca 2ϩ uptake than Rs-AFP2. Controlled Ca 2ϩ influx is believed to be essential for directing polar growth at the tip of fungal hyphae (35). For pollen tubes, which like fungal hyphae grow at their tip, it has been documented that various treatments resulting in elevated cytosolic Ca 2ϩ levels invariably lead to growth arrest (36). The three-dimensional structure of Rs-AFP1 has been studied by two-dimensional 1 H NMR, which has revealed that Rs-AFP1 consists of an ␣-helix (Asn-18 -Leu-28) and a triplestranded antiparallel ␤-sheet (␤-strand 1: Lys-2-Arg-6; ␤-strand 2: His-33-Tyr-38; ␤-strand 3: His-43-Pro-50) ( Fig. 6; Ref. 37). Meanwhile, the structure of Rs-AFP1 has been refined down to a root mean square deviation of 1.60 Å for all heavy atoms of the backbone, and the results of this refinement will be presented elsewhere. Since Rs-AFP1 is near-identical to Rs-AFP2, it is assumed that it adopts the same conformation. The spatial orientation of the residues affecting the antifungal activity of Rs-AFP2 upon substitution was analyzed using the high resolution structure of Rs-AFP1. According to the Rs-AFP1 model, all residues substituted in the present study do face outwards of the peptide backbone and are therefore unlikely to be essential for structure stabilization. The only exception is residue Ala-31, which is positioned at the interior face of the hairpin loop connecting the ␣-helix to ␤-strand 2 (Fig. 6). Substitution of Ala-31 by a bulky tryptophan residue in Rs-AFP2(A31W) most probably results in a conformational distortion, which might explain the drastic reduction of the antifungal activity of this variant. In addition, deletion of Pro-41, which adopts a cis-configuration in Rs-AFP1 as part of a type VI ␤-turn, is also likely to entail a distortion of at least the domain encompassing the second and third ␤-strand and the interconnecting type VI ␤-turn.
A graphical overview of the specific antifungal activity determined on F. culmorum of the different amino acid substitution variants when assayed in high ionic strength medium is provided in Fig. 7. When those residues affecting the antifungal activity are visualized on a three-dimensional model (Fig. 8), it becomes apparent that they all cluster into two adjacent sites. A first site is formed by the residues Tyr-38, Phe-40, Pro-41, Ala-42, Lys-44, and Ile-46. Except for Pro-41 and Lys-44, all those residues are highly hydrophobic. When Lys-44 was substituted by the neutral residue Gln, a substantial decrease of the antifungal potency was observed, suggesting that a positive charge within this predominantly hydrophobic cluster is important for the antifungal activity. This is further substantiated by the observation that the introduction of an additional positive charge within this site at position 39 resulted in enhanced antifungal activity in the presence of inorganic salts. The second site is formed by Thr-10, Ser-12, Leu-28, and Phe-49, which form a patch of contiguous residues despite their scattered positions along the Rs-AFP2 sequence (Fig. 7). Here again, introducing a positive charge within the cluster, namely at position 9, resulted in an enhanced antifungal potency in the high ionic strength medium.
The two regions important for the antifungal activity of Rs-AFP2 might constitute two sites contacting a single putative receptor. Alternatively, the presence of two sites could be indicative of two binding sites on each of two receptor molecules. The latter possibility has been proposed in a model for the interaction between the human growth hormone and its receptor (38). In the case of the human growth hormone, a mutational analysis has also revealed two domains that are involved in the interaction with the human growth hormone receptor. Each of the two domains interacts with a receptor molecule, entailing receptor dimerization, the initial trigger in the signal transduction pathway (39). The physiological meaning of the two functional sites of Rs-AFP2 will remain an open question until its putative receptor has been identified and characterized.
FIG. 8. Three-dimensional representation of Rs-AFP2 with indication of the residues affecting antifungal activity when substituted. The Rs-AFP1 molecule is represented in four orientations obtained by rotations of 90°about the vertical axis, with the ribbon presentation of the backbone at the top and the corresponding space-filling models at the bottom. The residues that caused a reduction of the antifungal activity in the medium SMFϩ by more than 4-fold when substituted are shown in dark blue (Thr-10, Leu-28, Tyr-38, Phe-40, Lys-44, Ile-46), and those which caused a reduction between 2-and 4-fold when substituted are indicated in light blue (Ser-12, Ala-42, Phe-49). The residues that enhanced the antifungal activity by more than 2-fold when substituted by arginine are marked in red (Gly-9, Val-39). Residue Ala-31, which is likely to entail major conformational changes in the backbone structure when substituted by tryptophan, and the deleted residue Pro-41 are indicated in green.