Differential Antifungal and Calcium Channel-Blocking Activity among Structurally Related Plant Defensins^,

Abstract

Plant defensins are a family of small Cys-rich antifungal proteins that play important roles in plant defense against invading fungi. Structures of several plant defensins share a Cys-stabilized α/β-motif. Structural determinants in plant defensins that govern their antifungal activity and the mechanisms by which they inhibit fungal growth remain unclear. Alfalfa (Medicago sativa) seed defensin, MsDef1, strongly inhibits the growth of Fusarium graminearum in vitro, and its antifungal activity is markedly reduced in the presence of Ca²⁺. By contrast, MtDef2 from Medicago truncatula, which shares 65% amino acid sequence identity with MsDef1, lacks antifungal activity against F. graminearum. Characterization of the in vitro antifungal activity of the chimeras containing portions of the MsDef1 and MtDef2 proteins shows that the major determinants of antifungal activity reside in the carboxy-terminal region (amino acids 31–45) of MsDef1. We further define the active site by demonstrating that the Arg at position 38 of MsDef1 is critical for its antifungal activity. Furthermore, we have found for the first time, to our knowledge, that MsDef1 blocks the mammalian L-type Ca²⁺ channel in a manner akin to a virally encoded and structurally unrelated antifungal toxin KP4 from Ustilago maydis, whereas structurally similar MtDef2 and the radish (Raphanus sativus) seed defensin Rs-AFP2 fail to block the L-type Ca²⁺ channel. From these results, we speculate that the two unrelated antifungal proteins, KP4 and MsDef1, have evolutionarily converged upon the same molecular target, whereas the two structurally related antifungal plant defensins, MtDef2 and Rs-AFP2, have diverged to attack different targets in fungi.

Plant defensins are small (45–54 amino acids) Cys-rich proteins implicated in the first-line host defense against fungal pathogens (Thomma et al., 2002). The tertiary structures of these proteins are quite similar and share a common Cys-stabilized α/β-motif composed of three antiparallel β-strands and one α-helix. This motif is also found in insect defensins and scorpion neurotoxins (Fontecilla-Camps, 1989; Bontems et al., 1991; Kobayashi et al., 1991). Despite their structural similarity, plant defensins are highly varied in their primary amino acid sequences, with only eight structure-stabilizing Cys residues in common (Thomma et al., 2002). The variation in the primary sequences may account for the different biological activities reported for plant defensins, including antifungal activity (Terras et al., 1995), antibacterial activity (Segura et al., 1998), proteinase activity (Wijaya et al., 2000), and α-amylase inhibitory activity (Bloch and Richardson, 1991).

Some plant defensins exhibit potent antifungal activity in vitro at micromolar concentrations against a broad spectrum of filamentous fungi. The morphogenic antifungal defensins reduce hyphal elongation and induce hyperbranching, whereas nonmorphogenic defensins reduce hyphal elongation without causing any morphological distortions (Broekaert et al., 1995; Thomma et al., 2003). Despite some progress made in the past few years, the structure-activity relationships and modes of action for most of the plant defensins remain unknown. Mutational analysis of the radish (Raphanus sativus) Rs-AFP2 has revealed that the amino acid residues important for antifungal activity are clustered into two adjacent sites. The first site is around the type VI β-turn connecting β-strands 2 and 3, and the second site is formed by residues on the loop connecting β-strand 1 and the α-helix and the contiguous residues on the α-helix and β-strand 3 (De Samblanx et al., 1997). Unlike the mammalian and insect defensins, antifungal plant defensins induce membrane permeabilization through specific interaction with high-affinity binding sites on fungal cells (Thevissen et al., 1997, 2000) but do not form ion-permeable pores in artificial lipid bilayers, nor do they change their electrical properties (Thevissen et al., 1996). When fungal hyphae are treated with Rs-AFP2 or the Dahlia merckii defensin Dm-AMP1, there is a rapid influx of Ca²⁺, efflux of K⁺, and alkalinization of the growth medium (Thevissen et al., 1996). How this interaction of defensin proteins with fungal hyphae generates plasma membrane ion fluxes leading to fungal growth inhibition remains unclear. The characterization of defensin-resistant mutants of unicellular and filamentous fungi has implicated a role for fungal sphingolipids and glucosylceramides in the mechanism of growth inhibition by these defensins (Thevissen et al., 2000, 2004; Ferket et al., 2003; Thomma et al., 2003). More recently, it has been shown that Rs-AFP2 interacts with fungal glucosylceramides in a first step leading to fungal growth arrest (Thevissen et al., 2004).

It is likely that not all plant defensins act by the same mode of action. For example, patch clamp experiments demonstrated that plant defensins isolated from maize (Zea mays) inhibit sodium currents in a rat tumor cell line. However, their in vitro antifungal activity was not reported (Kushmerick et al., 1998). Some scorpion neurotoxins are structurally related to defensins and are known to block potassium channels (Garcia et al., 2001). Based on surface topology similarities with potassium channel blockers, a similar mode of action has been proposed for a pea (Pisum sativum) seed defensin Psd1 (Almeida et al., 2002).

We have previously reported the isolation and characterization of the broad-spectrum antifungal defensin (MsDef1) from alfalfa (Medicago sativa) seed previously referred to as AlfAFP (Gao et al., 2000). This protein was found to inhibit the hyphal elongation of the fungal pathogen Fusarium graminearum in a dose-dependent manner, causing a hyperbranching phenotype. We demonstrate here that the carboxy-terminal region (residues 31–45) of MsDef1 contains major determinants for antifungal activity with the N-terminal region (residues 1–15) contributing to the antifungal activity in a relatively minor way. In addition, we show that Arg at position 38 is critical for the antifungal activity of MsDef1. This residue lies in a position homologous to the active site of the known Ca²⁺ channel blocker KP4 and the Na⁺ channel blocker scorpion toxin AaHII. Furthermore, we show for the first time, to our knowledge, that a plant defensin, MsDef1, selectively blocks the mammalian L-type Ca²⁺ channel in a manner similar to that of KP4, a structurally unique, virally encoded killer toxin from the P4 strain of the corn smut fungus, Ustilago maydis. However, two other structurally similar defensins, MtDef2 and Rs-AFP2, do not block the L-type Ca²⁺ channel. The blockage was found to be very strong (up to 90% blockage) and highly specific for the L-type channel. Finally, we report functional homology between MsDef1 and a known Ca²⁺ channel blocker, KP4, suggesting a common mode of action in fungi.

RESULTS

Alfalfa Defensins, MsDef1 and MtDef2, Differ in Their Antifungal Activity

The cloning and sequence analysis of the cDNA encoding the alfalfa seed defensin MsDef1 has been reported previously (Gao et al., 2000). A cDNA clone encoding MtDef2 was cloned and sequenced as described in “Materials and Methods.” The amino acid sequences of MsDef1 and MtDef2 are 65% identical and carry a net positive charge of +3 and −1, respectively (Fig. 1). The in vitro antifungal activities of these two proteins against a fungal pathogen, F. graminearum, were determined and compared with the antifungal activities of a previously characterized radish seed defensin, Rs-AFP2 (Terras et al., 1995), and the virally encoded killer toxin KP4 from U. maydis (Young, 1987; Park et al., 1994; Fig. 2; Table I). MsDef1, Rs-AFP2, and KP4 exhibited strong dose-dependent antifungal activity against this fungus, whereas MtDef2 did not. Although less potent, MsDef1 and KP4 induced more pronounced hyperbranching of the fungal hyphae than Rs-AFP2 (Fig. 2B). It should be noted that the most dramatic effect of these defensins is an altered growth pattern or hyperbranching that is accompanied by inhibition of hyphal elongation.

Figure 1.

Sequence comparison of Rs-AFP2 and Medicago defensins. Significant differences in residues between the two Medicago defensins are shown in boldface. The charge of each protein is shown in parentheses. Symbols represent approximate positions of the predicted α-helix (spiral) and β-sheets (arrows).

Figure 2.

Comparison between the effects of KP4, MsDef1, Rs-AFP2, and MtDef2 on F. graminearum. A, MsDef1, KP4, and Rs-AFP2 all exhibit a dose-dependent inhibition of hyphae. Assays were conducted in low ionic strength synthetic fungal media inoculated with fungal spores. Micrographs prepared after a 16-h incubation. B, Hyperbranching effect on Fusarium. In vitro assays were prepared in the same manner as described in A, with 25 μg mL⁻¹ of antifungal protein. The average number of hyphal buds per germline was determined by counting 50 germlings after 9 h of incubation. There is a marked difference in the number of buds among the proteins tested, with MsDef1 and KP4 inducing the most buds, Rs-AFP2 causing a less extreme effect, and MtDef2 causing no hyperbranching.

Table I.

IC₅₀ values for peptides

Antifungal Protein	IC50	IC50
		F. graminearum	N. crassa
	μg mL−1	μg mL−1
MsDef1	6–12	1–3
MtDef2	>100	25–50
Rs-AFP2	1–3	<1
KP4	>100	50–100
Def1-2C1	>100	25–50
Def1-2C2	6–12	1–3
Def1-2C3	>100	25–50
Def1-2C4	6–12	1–3
Def1-2C5	6–12	1–3
Def1-2C6	>100	25–50
Def2-Q39R	>100	1–3
Def1-R38Q	>100	25–50

Antifungal activity of defensins. The growth inhibition and hyperbranching induced by defensins were quantified. The concentration of defensins required to inhibit 50% of the overall growth (IC₅₀) was determined spectrophotometrically and confirmed visually. Fungal spores were grown in synthetic media supplemented with 2-fold serial dilutions of defensins. The absorbance at 595 nm was measured after 48 h.

C-Terminal Region (Residues 31–45) of MsDef1 Is Important for Its Antifungal Activity

In order to identify the regions of MsDef1 molecules that are important for antifungal activity and for induction of hyperbranching phenotype in the fungus, chimeric defensins consisting of portions of MsDef1 and MtDef2 proteins were tested for their antifungal activity and ability to induce hyperbranching in F. graminearum. MsDef1 and MtDef2 were divided into three regions of similar length, based upon the secondary structural elements predicted by sequence alignments with plant defensins whose three-dimensional (3D) structures have been determined (Fant et al., 1998; Almeida et al., 2002; Lay et al., 2003). Chimeric proteins corresponding to all six possible combinations (Fig. 3), termed Def1-2C1 through Def1-2C6, were obtained by expressing the synthetic genes encoding these proteins in Pichia pastoris. All six chimeric proteins were screened for antifungal activity against F. graminearum and Neurospora crassa, along with MsDef1 and MtDef2 proteins. Antifungal activity was assessed both by measuring hyphal growth inhibition after 16 h of exposure to the proteins and by determining the degree of hyperbranching after 9 h of exposure to the proteins.

Figure 3.

Antifungal activity of the chimeric Def1/Def2 defensins on F. graminearum. MsDef1 and MtDef2 were divided into three regions of similar length, and chimeric proteins corresponding to all six possible combinations, termed Def1-2C1 through Def1-2C6, were obtained by expressing the synthetic genes encoding these proteins using a P. pastoris expression system. All six chimeric proteins were screened for antifungal activity against F. graminearum and N. crassa (data not shown), along with MsDef1 and MtDef2 proteins. Antifungal activity was assessed by measuring both hyphal growth inhibition after 16 h of exposure to the proteins and determining the degree of hyperbranching after 9 h of exposure to the proteins (Table II).

As shown in Figure 3 and Table II, the C-terminal sequence (residues 31–45) of MsDef1 contains the major determinants of the in vitro antifungal activity since Def1-2C2 containing this sequence was nearly as active as MsDef1, whereas Def1-2C3 containing the corresponding sequence of MtDef2 was inactive against F. graminearum. The N-terminal sequence (residues 1–15) of MsDef1, however, does contribute somewhat to the antifungal activity of the C terminus since Def1-2C4 is more potent than either Def1-2C2 or Def1-2C5. As shown in Figure 3, both the N-terminal (residues 1–15) and C-terminal (residues 31–45) sequences of MsDef1 are required to observe the same degree of hyperbranching as that observed for the wild-type MsDef1. The middle region of the defensin (residues 16–30) apparently does not contain significant determinants of antifungal activity. Similar results were obtained when the chimeric defensins were tested on a more sensitive test fungus, N. crassa. It should be noted that, at concentrations higher than 12 μg mL⁻¹, all proteins, including Def1-2C1, Def1-2C3, Def1-2C6, and MtDef2, showed modest antifungal activity against N. crassa.

Table II.

Quantitation of defensin-induced hyperbranching in F. graminearum

Protein	No Peptides	MsDef1	MtDef2	Rs-AFP2	KP4	C1	C2	C3	C4	C5	C6
Ave	2.0	4.6	2.0	3.0	3.6	2.0	3.1	2.1	3.8	4.1	2.1
StD	0.0	1.3	0.1	1.0	0.9	0.1	1.0	0.2	1.0	1.2	0.2

The degree of hyperbranching was determined by counting the number of hyphal buds 8 h after incubation with defensin. Numbers in the table are averages of at least 50 spores per treatment.

Arg-38 Is Important for Antifungal Activity of MsDef1

In an effort to further define the region of MsDef1 conferring antifungal activity, we performed site-directed mutagenesis on MsDef1 and expressed the variant protein in P. pastoris. The antifungal activity of the purified protein was compared with that of MsDef1. We selected Arg-38 as a candidate for mutagenesis for the following reasons: first, the residue lies in the motif of MsDef1 most similar to the active site of the ion channel blocker KP4 and scorpion toxin AaHII (Fontecilla-Camps, 1989; Gu et al., 1995). Both proteins have been shown to have a basic residue at the base of the β2-β3 loop that is critical for its antifungal activity—Lys-42 for KP4 and Lys-58 for AaHII (Fontecilla-Camps, 1989; Gu et al., 1994; Gage et al., 2001). Second, Arg-38 lies in a homologous region shown to influence the antifungal activity of Rs-AFP2 (De Samblanx et al., 1997). Third, we noticed that Arg-38 is replaced by Gln in MtDef2. We suspected that the difference in charge at this position might account for a large difference in the antifungal activities of these two proteins (Table I). Therefore, we made a single amino acid substitution to both the MsDef1 protein (R38Q) and the MtDef2 protein (Q39R) and compared their antifungal activities against N. crassa and F. graminearum. We found that, against N. crassa hyphae, the Def2-Q39R was vastly more potent than the wild-type MtDef2 protein (Fig. 4; Table I) and almost identical to that of MsDef1. Conversely, the Def1-R38Q variant showed a dramatic decrease in antifungal activity compared to the wild-type MsDef1 protein and had similar antifungal activity as wild-type MtDef2. Therefore, the exchange of a single amino acid between MsDef1 and MtDef2 was able to switch the antifungal activity of the two defensins. Similar results were obtained against F. graminearum, with the exception that the Def2-Q39R mutant was not appreciably more potent than the wild-type defensin. It is likely that the molecular target of MsDef1 in this fungus is poorly recognized by Def2-Q39R and that additional residues in proximity to Arg-38 may be important for full recognition.

Figure 4.

Arg-38 of MsDef1 is important for antifungal activity. Spores of both fungi were allowed to germinate and grow in synthetic fungal media supplemented with 12 μg mL⁻¹ of the indicated defensin for N. crassa and 25 μg mL⁻¹ defensins for F. graminearum. Growth inhibition and hyperbranching were determined as described in “Materials and Methods.”

Antifungal Activity of MsDef1, Like That of KP4, Is Strongly Abrogated by Exogenous Ca²⁺

It has been shown previously that the antifungal activity of KP4 is specifically abrogated by exogenously added Ca²⁺ (Gu et al., 1995). Similarly, the antifungal activity of some plant defensins is significantly reduced when the cationic strength of the fungal growth medium is increased (Terras et al., 1992, 1993). Therefore, we tested the effects of Ca²⁺ on the in vitro antifungal activity of MsDef1 and Rs-AFP2 and compared them with that of KP4. As shown in Figure 5, a relatively low concentration of Ca²⁺ abrogates KP4 and MsDef1 activity against F. graminearum but does not affect the activity of Rs-AFP2. As little as 0.5 mm Ca²⁺ reduced the antifungal activity of KP4 and MsDef1, with total abrogation occurring at 2 mm Ca²⁺. The antifungal activity of KP4 was partially reduced by 2 mm Mg²⁺, but more than 5 mm Mg²⁺ was required for complete abrogation. With MsDef1, no reduction in antifungal activity was observed until more than 4 mm Mg²⁺ was added. For both KP4 and MsDef1, more than 25 mm Na⁺ or 25 mm K⁺ is required to reduce their antifungal activity and 50 mm is needed for complete abrogation. In terms of ionic strength, 3-fold more Mg²⁺ and 10-fold more K⁺ or Na⁺ are required to cause the same degree of reduction of antifungal activity of MsDef1 as Ca²⁺. Therefore, the in vitro antifungal activity of MsDef1, like that of KP4, is particularly sensitive to exogenously added Ca²⁺. Because the abrogation of MsDef1 and KP4 antifungal activity is metal specific and not simply due to ionic effects, it is likely that Ca²⁺ is involved in the mode of action of these proteins. The relative insensitivity of Rs-AFP2 to cations may be due to the fact that it binds to a different target than MsDef1 and KP4 or that it binds to the same target with significantly higher affinity, thereby making it harder to be displaced by Ca²⁺. As shown below, it seems more likely that the former possibility is true.

Figure 5.

Abrogation of the antifungal activity of MsDef1 and KP4 by Ca²⁺. Shown here is an example of one of the tested metals and concentrations; 50 μg mL⁻¹ KP4 and 5 μg mL⁻¹ MsDef1 and Rs-AFP2 were used in this experiment.

MsDef1, but Not MtDef2 and Rs-AFP2, Blocks the Mammalian L-Type Ca²⁺ Channel

It has been demonstrated previously that KP4 inhibits Ca²⁺ uptake in fungal cells (Gage et al., 2001). Rather unexpectedly, KP4 was found to specifically block the L-type voltage-gated Ca²⁺ channels in a weakly voltage-dependent fashion (Gu et al., 1995; Gage et al., 2002). Since the antifungal activity of MsDef1 closely parallels that of KP4, we tested MsDef1 for its ability to block a Ca²⁺ channel in mammalian cells. The activity of MsDef1 on Ba²⁺ currents conducted by three distinct voltage-gated Ca²⁺ channels, Ca_v2.1 (De Weille et al., 1991), Ca_v1.2 (Snutch et al., 1991), and Ca_v2.3 (Soong et al., 1993) from rat brain coexpressed with auxiliary subunits β_1b (Pragnell et al., 1991) and α₂δ (Ellis et al., 1988) in tsA-201 cells, was assayed as described previously (Gage et al., 2002). As shown in Figure 6A, 10 μm MsDef1 blocks approximately 90% of the Ca²⁺ current through the Ca_v1.2 (L-type) channel, with the maximum inhibition occurring after exposing the cells to the defensin for approximately 13 min. The block by 2 μm MsDef1 developed slowly over several minutes, such that 0.56% ± 0.03% of current remained at equilibrium (n = 3; Fig. 6B). A lower concentration was used in order to ascertain whether MsDef1 affected the voltage dependency of the current. Figure 6C shows the current-voltage relationship of Ca_v1.2 in the presence or absence of 2 μm MsDef1. Cells expressing Ca_v1.2, as described above, were held at −60 mV and depolarized to the indicated voltage for 100 ms before (black circles) or after (white circles) equilibration in 2 μm MsDef1. While MsDef1 decreases peak current, it does not appreciably shift the current-voltage relationship of Ca_v1.2 (Fig. 6C). MsDef1 did not block either the Ca_v2.1 or the Ca_v2.3 channel (Fig. 6, D and E). Surprisingly, Rs-AFP2 failed to block any of the three Ca²⁺ channels despite sharing a 3D structure similar to MsDef1 and having potent activity against F. graminearum (Fig. 6, F and G). As expected, MtDef2, which is closely related to MsDef1 but lacks in vitro antifungal activity against F. graminearum, also failed to block the L-type Ca²⁺ channel (Fig. 6H).

Figure 6.

MsDef1, but not Rs-AFP2 or MtDef2, selectively blocks L-type Ca²⁺ channels. A, Time dependency of the MsDef1 block of Ca_v1.2 channels. As shown here, 10 μm MsDef1 blocks approximately 90% of the Ca²⁺ current, with the maximum inhibition occurring after exposing the cells to the defensin for approximately 13 min. B, Block by 2 μm MsDef1 developed slowly, over several minutes, such that 0.56% ± 0.03% of current remained at equilibrium (n = 3). C, Current-voltage relationship of Ca_v1.2 in the presence and absence of 2 μm MsDef1. Cells expressing Ca_v1.2, as described above, were held at −60 mV and depolarized to the indicated voltage for 100 ms before (black circles) or after (white circles) equilibration in 2 μm MsDef1. While MsDef1 decreases peak current, it does not appreciably shift the current-voltage relationship of Ca_v1.2. D and E, MsDef1 was applied to the non-L-type channels Ca_v2.1 and Ca_v 2.3, respectively, as in B, except that the holding potential was −80 mV for Ca_v 2.1 and −100 mV for Ca_v 2.3. No inhibition of either Ca_v2.1 or Ca_v 2.3 by MsDef1 was detected. F, Whole-cell voltage clamp of tsA-201 cells expressing Ca_v1.2 channels, as in B, before (control) and several minutes after the application of 10 μm Rs-AFP2. G, The fraction of control current remaining after several minutes of perfusion with Rs-AFP2 in tsA-201 cells expressing Ca_v1.2, Ca_v2.1, or Ca_v2.3 channels. Ba²⁺ currents in these channels were elicited using the protocol described in B and D. The defensin Rs-AFP2 did not inhibit current conducted by any of these voltage-gated Ca²⁺ channels (values are means ± se; n = 3). H, L-type Ca_v1.2 channels are not blocked by MtDef2. The cell was held at −60 mm and pulsed to +10 mV before (control) or 5 min after initiation of perfusion with 10 μm MtDef2.

⁴⁵Ca²⁺ Uptake in N. crassa Hyphae

There have been several previous studies that attempted to measure Ca²⁺ flux across fungal membranes in response to defensins and KP4 (Thevissen et al., 1996, 1999; De Samblanx et al., 1997; Gage et al., 2001). We tested the ability of MsDef1, Rs-AFP2, and known Ca²⁺channel blockers to affect ⁴⁵Ca flux in N. crassa hyphae. Briefly, N. crassa spores were allowed to germinate in synthetic fungal media for 8 h. The cultures were then supplemented with 1 μCi mL⁻¹ of ⁴⁵CaCl₂. After a 20-min incubation, samples were collected and filtered onto Whatman filter paper, washed extensively with unlabeled CaCl₂, and counted in a liquid scintillation counter. We found that 25 μg mL⁻¹ of both MsDef1 and Rs-AFP2 caused an influx of Ca²⁺ (Table III). This result is consistent with what was reported previously for Rs-AFP2 (Thevissen et al., 1996). However, we found that four known Ca²⁺ channel blockers—cadmium, lanthanum, gadolinium, and KP4—also caused dramatic influx of Ca²⁺. Therefore, while this experimental approach is conceptually simple, it is apparently not adequate for measuring the activity of Ca²⁺ channel blockers. Indeed, this potential problem has been noted in a previous publication (Corzo and Sanders, 1992).

Table III.

⁴⁵Ca²⁺ uptake experiments in N. crassa

Treatment	Relative Influx	No. Experiments
MsDef1	4.4 ± 2.2	6
Rs-AFP2	14.8 ± 5.2	3
KP4	1.8 ± 0.1	3
Cd2+	3.1 ± 1.5	3
La3+	153.5 ± 35.2	4
Gd3+	102.7 ± 38.8	5

Spores were allowed to germinate for 8 h in synthetic fungal media before the addition of 1 μCi mL⁻¹ of ⁴⁵CaCl₂. MsDef1 and Rs-AFP2 were used at a concentration of 25 μg mL⁻¹; 50 μg mL⁻¹ KP4 were tested along with 1 mm of cadmium, lanthanum, and gadolinium. After 20 min, the hyphae were collected and vacuum filtered onto Whatman 5 filter paper, washed with 10 mL of 10 mm CaCl₂, and counted for ⁴⁵Ca. The relative influx was computed by taking the average cpm and dividing by the cpm of the control (untreated).

DISCUSSION

Plant defensins possess potent and broad-spectrum growth-inhibitory activity against fungi. Our findings shed light on the structure-activity relationships and modes of action of defensins from alfalfa. Our studies on the MsDef1 and MtDef2 chimeric defensins have shown that the C-terminal amino acid residues 31 to 45 in MsDef1 are important for antifungal activity. The comparison of this C-terminal region in these two defensins indicates the presence of four positively charged amino acids (Lys or Arg) in MsDef1 but none in MtDef2. Our analysis of the defensin chimeras also indicates that the N-terminal region (residues 1–15) contributes to the antifungal activity of MsDef1. This region is remarkably conserved in both proteins, with only two amino acid differences at positions 5 and 9. It is thus likely that the presence of Asn-5 and Lys-9 in MsDef1 contributes to the overall antifungal activity of MsDef1. Our analysis of the defensin chimeras indicates that the loop connecting β-strand 2 and β-strand 3, as well as β-strand 3, are the important secondary structure elements for the antifungal activity of MsDef1.

The structural homology of MsDef1 with the known voltage-gated Ca²⁺ channel blocker KP4 and the Na⁺ channel blocker scorpion toxin AaHII provided a guide for extending mutagenesis studies and, more precisely, defining the active site. All three of these proteins have common structural features, suggesting the active site lies near the β2-β3 loop. Like KP4 and scorpion toxin AaHII, this loop is extremely basic and stabilized by a disulfide bond with the C terminus. In the scorpion toxin, modification of Lys-58 at the base of the loop inactivates the toxin (Fontecilla-Camps, 1989). KP4 also has a Lys at the base of this loop (K42). When this residue was changed to a Gln, KP4 exhibited a 90% decrease in antifungal activity (Gage et al., 2001). MsDef1, like KP4 and scorpion toxin, has a basic residue at this position of the protein (Arg-38). Furthermore, MtDef2, which has limited antifungal activity, has Gln at this position. Because of the homology to MsDef1, we selected the Arg-38 for mutagenesis. As shown in Figure 4, this one residue was sufficient to reverse the antifungal activity of the mutant proteins when compared to the wild-type proteins. These results further demonstrate functional similarity between MsDef1 and the Ca²⁺ channel blocker KP4.

The results presented here further demonstrate that there is a great deal of mechanistic similarity between MsDef1 and the structurally different antifungal protein KP4 (Fig. 7). The data suggest that, although antifungal potency of these proteins differs significantly, their mechanism of action is similar. We show for the first time, to our knowledge, that a plant defensin blocks the L-type Ca²⁺ channel in mammalian cells. Like KP4 (Gu et al., 1995; Gage et al., 2002), MsDef1 is a potent inhibitor of the Ca_v1.2 (L-type) channel but has little or no effect on the Ca_v2.3 or Ca_v2.1 Ca²⁺ channels. Interestingly, MsDef1 blocks up to 90% of the L-type Ca²⁺ channel activity (Fig. 6A), whereas KP4 blocks only 60% of the L-type Ca²⁺ channel activity (Gu et al., 1995; Gage et al., 2002). In addition, MsDef1 takes approximately 13 min to reach equilibrium and is very reminiscent of the time-dependent effects of calciseptine (Teramoto et al., 1996). Like KP4, this block does not seem to change the voltage dependency of the channel appreciably (Fig. 6C). These observations, along with the abrogation of antifungal activity by Ca²⁺, suggest that MsDef1 binds to the extracellular side of the Ca_v1.2 pore region much like the blockage of K⁺ channels by charybdotoxin (MacKinnon and Miller, 1989) or the blockage of Na⁺ channels by tetrodotoxin (Terlau et al., 1991).

Figure 7.

Three-dimensional structures of the plant defensin Rs-AFP1 (1AYJ; Fant et al., 1998), the scorpion toxin AaHII (1PTX; Housset et al., 1994), and the fungal toxin KP4 (1KPT; Gu et al., 1995). In all three figures, the Cys side chains are represented in ball and stick, while the Arg and Lys residues are shown as blue stick models. In the diagram of Rs-AFP1, the colors of the secondary structural elements correspond to the regions selected for MsDef1 hybrid analysis. Orange, purple, and green indicate the N-terminal (residues 1–15), middle (residues 16–30), and C-terminal (residues 31–45) portions, respectively. For scorpion toxin and KP4, the ribbon diagrams are colored in a gradient from red to purple as the protein is traced from the N to the C termini.

In order to test whether other morphogenic defensins share this property with MsDef1, Rs-AFP2 was chosen since its 3D structure is similar to that of MsDef1 but differs substantially in its primary amino acid sequence (Fant et al., 1998). Rs-AFP2 is a more potent growth inhibitor of F. graminearum and N. crassa than MsDef1, but it induces less pronounced hyperbranching in these fungi. Rs-AFP2 failed to block any of the three Ca²⁺ channels (Fig. 6, F and G), indicating different modes of action for these two structurally related defensins (Lay et al., 2003). The fact that MsDef1, like KP4, specifically blocks the L-type Ca²⁺ channel and that its antifungal activity is abrogated by Ca²⁺ makes it likely that MsDef1 targets fungal Ca²⁺ channels. Although there may be an alternative mode of action, the data presented here are consistent with the notion that MsDef1 blocks a specific fungal Ca²⁺ channel. While MsDef1 has not been directly shown to block a fungal Ca²⁺ channel, it should be noted that there is no known example of an ion channel blocker that has an unrelated function in its active biological system. Furthermore, it should be noted that disruption of fungal Ca²⁺ gradients is known to cause hyperbranching (Jackson and Heath, 1993) and that the efficacy of these proteins on L-type Ca²⁺ channels correlates well with their hyperbranching effects.

It is reasonable to suggest that MsDef1 acts via disruption of a Ca²⁺ gradient, since Ca²⁺ is a ubiquitous signaling molecule that plays important roles in the life cycle of fungi. In fungi, Ca²⁺ is involved in, but not limited to, bud formation (Davis, 1995), hyphal elongation (Jackson and Heath, 1993), and cAMP regulation (Iida et al., 1990). This Ca²⁺ gradient is maintained through a series of Ca²⁺ channels, antiporters, and pumps (Cunningham and Fink, 1994; Tsien et al., 1995). Hyphal tip growth is a highly dynamic and complex process that involves control of localized synthesis and expansion of the growing tip. This process is controlled by a gradient in the cytosolic Ca²⁺ generated by tip-localized Ca²⁺ channels (Tsien and Tsien, 1990). Indeed, disruption of this tip gradient has been shown to cause hyperbranching in growing hyphae much like that observed with KP4 and MsDef1 (Jackson and Heath, 1993).

If MsDef1 causes hyphal growth defects by blocking the uptake of Ca²⁺, then N. crassa growth should also be inhibited by EGTA and lanthanum, a known Ca²⁺ channel blocker. When N. crassa spores were grown in synthetic media supplemented with EGTA or lanthanum, fungal growth was inhibited and hyperbranching was induced in a manner resembling that of MsDef1 (see Supplemental Figs. 1 and 2, available at www.plantphysiol.org). Concentrations of EGTA over 100 μm greatly inhibited the growth of N. crassa, and concentrations over 500 μm inhibited the growth of F. graminearum. Growth inhibition of U. maydis by EGTA was previously demonstrated to be similar to that caused by KP4 (Gage et al., 2002). This finding is consistent with the notion that defensins may be inhibiting normal hyphal growth by disrupting Ca²⁺ transport.

It is possible that the antifungal effects of MsDef1 result from the disruption of Ca²⁺ gradients required for filamentous growth and budding rather than inhibition of nutritional uptake of Ca²⁺ (Jackson and Heath, 1993). It has been reported that mutations in either adenylyl cyclase (Gold et al., 1994) or cAMP-dependent protein kinase (Durrenberger et al., 1998) can affect the control of filamentous growth versus budding in fungi. The addition of the secondary messenger, cAMP, to the growth media can revert the mutant phenotype to a wild-type phenotype. In the case of KP4, exogenously added cAMP was found to rescue the growth inhibition induced by this Ca²⁺ channel blocker (Gage et al., 2001). Because of the functional similarity of MsDef1 with KP4, we determined if abrogation of MsDef1-induced antifungal activity occurred with cAMP treatment. As in the case with KP4, cAMP reduced the antifungal activity of MsDef1 in a dose-dependent manner for both N. crassa and F. graminearum. Spores of both fungi were germinated in the presence of inhibitory concentrations of MsDef1 (2 μg mL⁻¹ for N. crassa and 25 μg mL⁻¹ for F. graminearum). cAMP at concentrations above 10 mm was able to reduce the antifungal effects of MsDef1 in a dose-dependent manner (see Supplemental Fig. 3). This result suggests a link between growth inhibition and fungal Ca²⁺ signaling (Gage et al., 2001).

In N. crassa, there are three distinct Ca²⁺ signal transduction pathways based on the unique Ca²⁺ signatures associated with mechanical perturbation, hypo-osmotic shock, and high external Ca²⁺. KP4 inhibits the intracellular Ca²⁺ responses to hypo-osmotic shock and high external Ca²⁺ but not to mechanical perturbation (Nelson et al., 2004). In addition, physiological evidence suggests the presence of two stretch-activated and two intracellular inositol-1,4,5-triphosphate-activated Ca²⁺ channels in N. crassa (Levina et al., 1995; Silverman-Gavrila and Lew, 2001, 2002).

The complete sequencing of the N. crassa genome has revealed the presence of three Ca²⁺ channel genes (Galagan et al., 2003). Three new Ca²⁺ channel proteins have been identified in Neurospora (Borkovich et al., 2004). These three proteins have close homologs to the three Ca²⁺ channel proteins in Saccharomyces cerevisiae—Mid1p, Cch1p, and Yvc1p (Fischer et al., 1997; Paidhungat and Garrett, 1997; Muller et al., 2001). The sequence of CCH1 is similar to the α₁-subunit of animal voltage-gated Ca²⁺ channels. Yvc1p is a vacuolar voltage-gated Ca²⁺ channel (Palmer et al., 2001). MID1 does not have any sequence similarity to known ion channels but has been reported to be a stretch-activated cation channel (Kanzaki et al., 1999).

In order to test MsDef1 directly for its Ca²⁺ channel-blocking activity in N. crassa, it will be necessary to conduct electrophysiological studies with specific Ca²⁺ channels in the same manner as reported for the mammalian Ca²⁺ channels. Currently, it has been difficult to directly measure defensin-induced Ca²⁺ flux in a fungal system. Previous studies have looked at ⁴⁵Ca in N. crassa hyphae treated with Rs-AFP2 and reported a rapid influx (Thevissen et al., 1996; De Samblanx et al., 1997). In an effort to determine if the treatment of the fungus with MsDef1 leads to significant changes in Ca²⁺ flux across the plasma membrane, we performed a ⁴⁵Ca²⁺ uptake assay using N. crassa hyphae. All defensins tested caused a strong influx of Ca²⁺ similar to that previously reported for Rs-AFP2 (Table III; Thevissen et al., 1996). While this result appears to suggest that defensins act as Ca²⁺channel activators, hyphae treated with the known Ca²⁺ channel blockers—cadmium, gadolinium, lanthanum, and KP4—also caused a strong influx of Ca²⁺. As noted in previous publications, data generated from ⁴⁵Ca²⁺ uptake experiments must be interpreted with extreme caution and may not necessarily reflect actual Ca²⁺ uptake (Corzo and Sanders, 1992). Due to the complexity of Ca²⁺ regulation in fungal hyphae, it is difficult to assess the Ca²⁺ channel-blocking activity of defensins with this simplistic technique. With the availability of the Ca²⁺ channel gene sequences in N. crassa, it may now be possible to test MsDef1 for its ability to block these channels individually using the same approach we report for the mammalian channels.

MATERIALS AND METHODS

Cloning and Sequence Analysis of Medicago Def1 and Def2 Genes

The cloning and nucleotide sequence of a full-length cDNA clone encoding MsDef1 has been published previously (Gao et al., 2000). A search of The Institute of Genomic Research M. truncatula Gene Index (MtGI) using the MsDef1 sequence as a query yielded one singleton expressed sequence tag clone (TC50237) of 483 bp with 75% identity at the nucleotide sequence level. This cDNA clone encoding MtDef2 was generously provided by The Samuel Roberts Noble Foundation (Ardmore, OK) and sequenced (J.N. Hanks, A.K. Snyder, M.A. Graham, R.K. Shah, L.A. Blaylock, M.J. Harrison, and D.M. Shah, unpublished data).

The sequence for Rs-AFP2 was published previously (Terras et al., 1992). We obtained a synthetic gene encoding the mature Rs-AFP2 protein from Integrated DNA Technologies (Coralville, IA) and cloned it into the Pichia pastoris expression vector as described below.

Expression of Defensins in Pichia pastoris

All defensin proteins were expressed in the yeast Pichia pastoris. The pPIC9 vector (Invitrogen, Carlsbad, CA) allows the methanol-inducible expression of the recombinant protein in P. pastoris and its secretion using the α-factor secretion signal of Saccharomyces cerevisiae. The DNA sequences coding for the mature defensin sequence of MsDef1, MtDef2, and Rs-AFP2 were cloned in frame with the initiation codon of the signal sequence at the XhoI restriction site of pPIC9. The sequence encoding the last four amino acids of the α-factor signal protein sequence were not included in the expression construct. The plasmids were transformed in Escherichia coli DH5α. The resulting vector contained the coding region for the mature defensin sequence fused in frame with the α-factor signal sequence downstream of the P. pastoris alcohol oxidase promoter. The vector was then linearized by digestion with SalI and integrated into P. pastoris strain GS115 (Invitrogen) by electroporation. His⁺ transformants were selected by plating on minimal dextrose plates. Clones were cultured in buffered minimal glycerol media and induced with methanol. The presence of each defensin in the growth medium was confirmed by ELISA (Gao et al., 2000).

Design of Defensin Chimeras

Def1-2C5, Def1-2C6, Def1-R38Q, and Def2-Q39R expression vectors were prepared by site-directed mutagenesis of the MsDef1 and MtDef2 expression vectors. The mutagenesis was done by PCR using the QuickChange site-directed mutagenesis kit purchased from Stratagene (La Jolla, CA). This technique changed the Asn-5 residue and Lys-9 of MsDef1 to His and Thr, respectively, creating Def1-2C5. Similarly, the His-5 and Thr-9 residues of the MtDef2 construct were mutated to Asn and Lys to create Def1-2C6. The synthetic genes for Def1-2C1, Def1-2C2, Def1-2C3, and Def1-2C4 were obtained from MCLAB (San Francisco) and cloned into pPIC9 in a manner identical to that employed for MsDef1, MtDef2, and Rs-AFP2.

Expression and Purification of Defensins

Pichia cultures were grown overnight in buffered minimal glycerol media and then induced with methanol every 24 h, according to the manufacturer's directions (Invitrogen). The cultures were grown for 7 d at 29°C, and cells were removed by centrifugation at 2,000g for 15 min.

Defensins were purified from the growth medium by first dialyzing against 25 mm sodium acetate, pH 4.5, and then passing the dialyzate through CM-Sephadex C-25 cation-exchange resin (Amersham Biosciences, Piscataway, NJ) equilibrated with 25 mm sodium acetate, pH 4.5. Resin was extensively washed with binding buffer (25 mm sodium acetate, pH 4.5), and the bound protein was then eluted in 1 m NaCl, 50 mm Tris, pH 7.6. Fractions containing the protein were manually collected and analyzed by SDS-PAGE for the presence of the defensin. Fractions containing the defensin protein were concentrated using a Minitan II ultrafiltration system (Millipore, Bedford, MA) with a 3-kD cutoff membrane and dialyzed against 10 mm Tris, pH 7.6. Purity was assessed using Homogenous 20 SDS gels on a Phastgel system (Amersham Biosciences). The identity of the defensin was confirmed by MALDI-MS in the positive ion mode using the matrices α-cyano-4-hydroxy cinnamic acid (reflector mode) and sinapinic acid (linear mode). All defensins were found to be pure and have the predicted mass-to-charge ratio (data not shown).

The protein concentration for all expressed defensins was determined by bicinchoninic acid protein concentration assay using the test tube protocol provided by the manufacturer (Pierce, Rockford, IL).

Purification of KP4

KP4 was purified as reported previously (Gu et al., 1994, 1995). Briefly, the toxin was isolated from the supernatant of the KP4 toxin expressing strains of Ustilago maydis strain P4 using ion-exchange chromatography similar to that described above.

Antifungal Assays

The antifungal activity of MsDef1 and MtDef2 and chimeric defensins was measured in an in vitro assay using 96-well microtiter plates. Fifty microliters of each protein dilution were added to each well of the microtiter plate containing 50 μL of spore suspension prepared in 2× synthetic low-salt fungal medium at a concentration of 40,000 spores mL⁻¹ (Liang et al., 2001). Fusarium graminearum and Neurospora crassa spores were harvested from potato dextrose (Difco, Sparks, MD) and Vogel's medium N (Vogel, 1964) agar plates, respectively, by washing the plate with water. The fungal cultures were incubated at 24°C for 24 h. The IC₅₀ values were determined by microscopic analysis of hyphal growth after a 16-h incubation at room temperature and microspectrophotometrically after 48 h (Broekaert et al., 1990). Values were obtained from data collected in at least three independent experiments. It should be noted that accurate measurements of growth inhibition with N. crassa are difficult because of the aerial hyphae that develop. IC₅₀ values do not always indicate the degree of antifungal activity. Extreme hyperbranching can cause denser growth and thereby higher absorbance in this assay. The photographs were taken using an Olympus CK40 inverted microscope at 200× magnification using Kodak T-max 400 (Rochester, NY) film after 16 h of growth, unless otherwise noted. The degree of hyperbranching of F. graminearum hyphae was quantified by counting the number of hyphal buds visible on the germlings after 9 h of incubation with defensin proteins. More than 50 spores from each treatment were counted per experiment. The hyperbranching assay was repeated at least twice.

To test the growth-inhibitory effects of EGTA and lanthanum, spores were used to inoculate 100 μL of synthetic fungal medium in a 96-well microplate at a concentration of 2,000 spores per well (Broekaert et al., 1990). The cultures were supplemented with various concentrations of EGTA or lanthanum chloride. Growth was measured at the indicated times by determining optical density at 595 nm in a spectrophometer (Spectra Max; Molecular Devices, Sunnyvale, CA). Optical density readings were taken after 15, 37, and 48 h of growth at 25°C. The antifungal assay for cAMP abrogation was done similarly to that described above. For N. crassa, 2 μg mL⁻¹ of MsDef1 were added to the culture in addition to the indicated concentrations of cAMP.

Effect of Defensins on Mammalian Ca²⁺ Channels

The activity of MsDef1 and Rs-AFP2 on Ba²⁺ currents passing through three distinct Ca²⁺ channels was tested by an assay described previously (Gage et al., 2002). Briefly, whole-cell voltage clamp recording of Ba²⁺ currents in tsA-201 cells expressing the L-type channel Ca_v1.2 along with the β1b and α₂δ subunits was made. The cell was held at −60 mV, and current was elicited using 100-ms depolarizations to +10 mV at a frequency of 0.05 Hz. After a steady baseline of current was established, defensin protein was applied to cells via perfusion in the extracellular bath. The bath solution consisted of 150 mm Tris, 4 mm MgCl₂, and 10 mm BaCl_2, pH 7.3. The pipette solution consisted of 130 mm N-methyl, d-glucamine, 60 mm HEPES, 10 mm EGTA, 2 mm MgATP, and 1 mm MgCl₂, pH 7.3. Currents were recorded using an Axopatch 200B amplifier (Axon Instruments, Union City, CA). Defensins were applied in the extracellular bath at the indicated concentrations using an RSC 160 perfusion system (Bio-Logic, Claix, France). Data acquisition and analysis were performed using Clampex and ClampFit software (Axon, Raleigh, NC). All current recordings were leak subtracted using a standard P/N procedure and filtered at 1 kHz.

⁴⁵Ca²⁺ Uptake in N. crassa Hyphae

In order to measure the flux of ⁴⁵Ca²⁺ across fungal hyphae, N. crassa spores were used to inoculate cultures of synthetic fungal media at a concentration of 3 × 10⁵ spores mL⁻¹ in 50-mL Falcon tubes. The spores were allowed to germinate by incubating the cultures at 28°C for 8 h in a shaking incubator (200 rpm); 250-μL aliquots of the culture were prepared and added to 1.5-mL microcentrifuge tubes. The cultures were then supplemented with 1 μCi mL⁻¹ of ⁴⁵CaCl₂ and 25 μg mL⁻¹ of MsDef1, Rs-AFP2, KP4, or 1 mm gadolinium chloride, lanthanum chloride, or cadmium chloride. After a 20-min incubation at room temperature, the hyphae were collected and vacuum filtered onto Whatman 5 filter paper. Each filter was washed with 10 mL of 10 mm CaCl₂. The filter paper was then counted for ⁴⁵Ca in a Beckman LS 6500 liquid scintillation counter (Beckman Instruments, Fullerton, CA).

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession number MtDef2 AY313169.