INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Beticolins are non-host-specific toxins produced by the phytopathogenic fungus Cercospora beticola, which is responsible for the leaf spot disease on sugar beet. Beticolins form a family of nonpeptidic compounds (with 20 identified to date, named beticolin 0 to beticolin 19). They share the same polycyclic skeleton with a chlorine atom and partially hydrogenated anthraquinone and xanthone moieties (Fig. 1 A; see also Ducrot et al., 1994; Prangé et al., 1997, and references therein) but differ by isomeric configurations (ortho- or para-, Fig. 1 A) and by a variable residue (R in Fig. 1 A).

View larger version (27K): [in this window] [in a new window]   FIGURE 1   Structure of beticolins. (A) Structure of ortho- and para-beticolin. (B) B3 is a para-beticolin with a residue (R), which is C3H4O2 (Ducrot et al., 1994).

Beticolins are endowed with many cytotoxic effects. In plants, beticolins have been found to induce dramatic loss of solutes such as amino acids and -cyanin from root tissues (Schlösser, 1971; Macri and Vianello, 1979). These toxins inhibit ATP-dependent H⁺ transport in pea stem (Macri et al., 1983) and corn root (Blein et al., 1988). In tobacco cells, beticolins cause dissipation of transmembrane potential (Gapillout et al., 1996). In addition to their toxicity on plant plasma membrane, beticolins also exhibit an antibiotic activity (Schlösser, 1962). In vitro, beticolins display an O₂ scavenging activity at lower concentration than do vitamin E and Tiron (Rustérucci et al., 1996). Moreover, beticolins show an antiproliferative effect on rat adrenocortical cell lines and mouse tumor adrenocortical cells by modulating a step of steroid biosynthetic pathway (Ding et al., 1996).

Induced permeabilization of cell membranes is one of the most common killing mechanisms of molecules that have cytotoxic functions. For several peptidic or nonpeptidic toxins or antibiotics, formation of multimeric pores has been suggested to be responsible for membrane permeabilization and cell lysis (e.g., alamethicin (Woolley and Wallace, 1992; Sansom, 1993), gramicidin (Woolley and Wallace, 1992), amphotericin B (Ramos et al., 1996), flammutoxin (Tomita et al., 1998), lycotoxin (Yan and Adams, 1998), yeast killer toxins (Kagan, 1983); see Bernheimer and Rudy, 1986; Ojcius and Young, 1991; Marsh, 1996; and Mirzabekov et al., 1999, for reviews). The deleterious properties of beticolins could originate from a pore-forming feature of these fungal toxins. Recently we have shown that (ortho-)beticolin 0 can form ion-permeable channels at cytotoxic concentration (Goudet et al., 1998). However, the generalization of the channel formation to the whole beticolin family is still to be demonstrated. Furthermore, the secondary structures of the pore remain essentially unknown, and little is known about the mechanisms underlying this pore formation.

Here we report that (para-)beticolin 3 (B3) (see Fig. 1 B) is able to form voltage-independent and weakly selective ion channels in planar lipid bilayers. These channels present different conductance levels, which are multiples of each other. To understand the structures underlying this property, we compared the ionic selectivity of the first three conductance levels in different salt media. We also compared the pore size of the first three conductance levels by studying the effect of different nonelectrolytes of various sizes on beticolin conductances and analyzing the dependence of single-channel conductance on the hydrodynamic radius of nonelectrolytes. Considering the high similarity of both ionic selectivity and pore radius between the different conductance levels, a cluster organization of elementary B3 channels that open and close simultaneously is likely to explain the existence of the different conductance levels. Furthermore, with our results and with data from analysis of x-ray diffraction by B3 microcrystals, we propose a model for the 3D structure of elementary ion-conducting channel(s) formed by B3 cytotoxin.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Planar lipid bilayer

Planar bilayers were formed by painting a solution of synthetic 1-palmitoyl-2-oleoyl-phosphatidylethanolamine (POPE) and 1-palmitoyl-2-oleoyl-phosphatidylcholine (POPC) (Avanti Polar Lipids, Birmingham, AL), in a 1:1 molar ratio, dissolved in n-decane (10 mg·ml¹; Sigma) across a 0.2-mm-diameter hole made in polyvinylidenedifluoride septum separating two chambers of 2 ml (defined as cis and trans) (Mueller et al., 1962).

Currents through the bilayer were recorded with an Axopatch 200B (Axon Instruments, Foster City, CA) patch-clamp amplifier. Trans and cis chamber solutions were connected (through 3 M KCl/AgCl/Ag half-cells) to headstage and signal-ground inputs of the amplifier, respectively. Potential values were thus defined as trans chamber minus cis chamber voltage. Membrane formation was verified by monitoring membrane capacitance and resistance. Bilayers readily formed with a capacitance of 100-150 pF. For a period of at least 30 min before toxin addition, those membranes were stables and, in standard solution, had a conductance of less than 10 pS at voltages in the ±100 mV range. Recordings were sampled at 2 kHz and filtered at 50 Hz for further analysis using pClamp 6 (Axon instruments) and Origin 4.0 (Microcal Software, Northampton, MA) softwares.

Chemicals

The standard solution contained 200 mM KCl, 0.88 mM MgCl₂, 1 mM EDTA, and 10 mM HEPES-KOH (pH 7.4). The free Mg²⁺ concentration is estimated to be 10 µM. In standard experiments, both the cis and trans chambers contained the standard solution.

To determine the channel ionic selectivity, the reversal potential was measured in asymmetrical solutions of a fivefold XCl concentration gradient (500 mM/100 mM XCl, cis/trans, with X = K, Na, Li, NH₄, or tetraethylammonium, TEA), substituting KCl in the standard solutions. The cation/anion permeability ratio was then calculated using the Goldman-Hodgkin-Katz equation (Hille, 1992).

Beticolin 3 was extracted from C. beticola mycelium (strain CM) and purified as previously described (Milat et al., 1993). B3 was added in the stirred cis solution at a concentration of 10 µM.

Pore size determination

Pore size determination was performed using nonelectrolyte (NE) molecules of different hydrodynamic sizes (Carneiro et al., 1997; Krasilnikov et al., 1997, 1998; Kaulin et al., 1998; Krasilnikov et al., 1998). Briefly, 20% (w/v) of a specified NE was added to the standard solution in both chambers. The following NEs were used: glucose, fructose, sorbitol (Sigma), and different polyethylene glycols (PEGs): 300, 400, 600, 1000, 3000, and 8000 (Fluka). The conductivities of these solutions were measured with a conductimeter (CD16; Tacusel Electronique, Villeurbanne, France). Then the maximum pore radius was determined by measuring B3 single-channel conductances () in the different solutions. As described by Krasilnikov et al. (1997), the filling parameter () was calculated as follows:

(1)

Where _s and ₀ are the single-channel conductances, respectively, in the presence or absence of a given NE and _s and ₀ are the conductivities of the solutions. The relationship between  and the hydrodynamic radius of the NE allowed the determination of the maximum pore radius from the intercept between the decreasing linear portion of the relation (where NE is able to enter the channel and  decreases with increasing NE radius) and the quasihorizontal portion (where NE is unable to enter the pore and  is independent of the hydrodynamic radius).

All results from different experiments are expressed as mean ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Multiple conductance levels of B3 channels

A few minutes after the addition of 10 µM B3 to the cis medium, discrete jumps in the current to multiple levels were observed. Fig. 2 A shows an example of a 50-s-long current record displaying different amplitude levels. The corresponding amplitude histogram is shown in Fig. 2 B. The envelope of this histogram could be satisfactorily fitted by a sum of four Gaussian functions (dashed curve in Fig. 2 B). The four corresponding peaks (solid lines in Fig. 2 B) are centered on the four discrete current levels that can be distinguished in the record (four dashed lines in Fig. 2 A). The left-hand peak corresponded to the leak current (<1 pA) and was taken as the baseline. The other peaks were centered on the following values (relative to the baseline): 1.3, 3.1, and 6.0 pA (from left to right). In this example only four current levels could be detected. But whenever more peaks were observed in other experiments (not shown, but see Discussion), they were always centered on values in an approximately twofold geometric progression.

View larger version (32K): [in this window] [in a new window]   FIGURE 2   Beticolin 3 currents of different amplitudes through planar lipid bilayers. (A) Representative trace of channel activity recorded at +100 mV in standard symmetrical solutions after the addition of 10 µM B3 toxin to the cis solution. Experiments have been made using synthetic POPE/POPC lipid bilayers. Both cis and trans standard solutions contained 200 mM KCl, 0.88 mM MgCl2, 1 mM EDTA, and 10 mM HEPES-KOH (pH 7.4). The free Mg2+ concentration was estimated to be 10 µM. Data were recorded at 2 kHz and filtered at 50 Hz. The current of the closed (0) state is indicated by the solid line. The dashed lines figure the three different discrete current levels (marked by 1, 2, and 3, derived from analysis of the histogram shown in B). (B) Frequency histogram of current amplitude (data from A) and multi-Gaussian fit. (C) Mean current-voltage relationships for the three open states (, level 1; , level 2; , level 3) of B3 channels (n = 4). Solid lines are linear fits, of which the slope corresponds to the following conductance values: 16 ± 1 pS, 32 ± 2 pS, and 57 ± 2 pS, respectively, for levels 1, 2, and 3.

Current records similar to that presented in Fig. 2 A were measured at different voltages from 120 to +120 mV. From this we derived the current-voltage relation of the three conductance levels detected above. The three I-V curves were linear and crossed the origin (Fig. 2 C). The slope conductances were 16 ± 1, 32 ± 2, and 57 ± 2 pS (n = 4). The first three conductance levels elicited by B3 are referred to as 1 to 3. A fourth conductance level (4) was occasionally observed and had a conductance value of 111 ± 3 pS (n = 3).

Ionic selectivity of B3 current

The cation/anion selectivity of B3 channels was determined by measuring the reversal potential (E_rev) in a 5:1 (cis/trans) concentration gradient of chloride salts, KCl, NaCl, LiCl, NH₄Cl, or TEACl (Fig. 3). Under any condition, E_rev was approximately the same for the first three conductance levels, 1 to 3: the values were within 3 mV of the average given in each panel of Fig. 3. This finding clearly indicates a similar ionic selectivity for 1 to 3.

View larger version (16K): [in this window] [in a new window]   FIGURE 3   Similar reverse potential of the first three amplitude levels of B3 channels in different asymmetrical media. Current-voltage relationshipsof the multiple B3 conductance levels (, level 1; , level 2; , level 3) in different asymmetrical media. The trans medium contained 100 mM XCl (where X = K in A, Na in B, Li in C, NH4 in D, and TEA in E), whereas the cis medium contained 500 mM XCl. Both cis and trans medium also contained 0.88 mM MgCl2, 1 mM EDTA, and 10 mM HEPES-KOH (pH 7.4). The free Mg2+ concentration was estimated to be 10 µM. In each condition, reverse potentials of the different amplitude levels were very close. The data shown are representative of at least three experiments.

From the E_rev values obtained in asymmetrical KCl, NaCl, LiCl, NH₄Cl, or TEACl solutions, permeability ratios (P_X/P_Cl) were derived (Table 1), and the following selectivity sequence was deduced: Na⁺ > Li⁺ > K⁺ > NH₄⁺ > TEA⁺ > Cl.

Pore size determination

The maximum pore size was determined by analyzing B3-elicited conductances in solutions containing 20% (w/v) NEs with different hydrodynamic radii. These compounds decreased the electric conductivity of the solution (Table 2) and modified the  values elicited by B3. When plotted against the NE hydrodynamic radius (r), 1, 2, and 3 displayed a similar pattern of variation (Fig. 4). NEs with r smaller than 7 Å affected  values, while those with r greater than 8 Å did not (horizontal dashed lines in Fig. 4). These data suggest that any conductance level involved a similar pore radius in the range of 7-8 Å (marked by vertical dotted lines).

View larger version (21K): [in this window] [in a new window]   FIGURE 4   Similar dependence of the first three conductance levels of B3 channels on the hydrodynamic radii of nonelectrolytes (NEs) added to both sides of the bilayer. Chord conductance of single channels () versus NE radius (, level 1; , level 2; , level 3). Single-channel events were recorded at +100 mV in standard symmetrical solutions (as indicated in Fig. 2) containing 20% (w/v) of a test NE. Average conductance was calculated from more than 100 current transitions registered in each experimental condition, using Ohm's law. The hydrodynamic radii of NE were taken from Table 2. For the three different conductance levels a similar relation was observed. The horizontal dashed lines were plotted between experimental points corresponding to NE's radius in the 8-30.5-Å range, where  did not vary. The two vertical dotted lines indicated the minimum (~7 Å) and maximum (~8 Å) values of the apparent pore size of B3 channels.

Consideration of the filling parameter  (calculated according to Eq. 1) has been shown to allow more acute determination of the pore size (Krasilnikov et al. 1998). Consistent with our above findings, the dependence of  on r was similar for 1, 2, and 3 (Fig. 5). As described by Krasilnikov et al. (1998), an estimate for the pore radius was obtained from the intercept between the decreasing linear and the quasihorizontal portions of the relationship. Following this method, the estimated pore radius was close to 7.5 Å for all three conductance levels considered (Fig. 5).

View larger version (21K): [in this window] [in a new window]   FIGURE 5   Similar estimated pore size of the first three conductance levels of B3 channels. The filling parameter  derived from the first three conductance levels of B3 channels (from top to bottom: (A) , level 1; (B) , level 2; (C) , level 3) are plotted against the hydrodynamic radii of nonelectrolytes (NEs) added to both sides of the bilayer. The parameter  was calculated using Eq. 1 (see in Experimental Procedures) according to the conductance values from Fig. 4. In the three panels, linear regressions on data corresponding to NE's radius in the 8-30.5-Å range resulted in a quasihorizontal line. Other sloped linear regressions were made on data corresponding to NE's radius in the 3.7-7-Å range. The intersection between the former and latter linear regressions gave an estimate of the maximum pore radius of B3 channels. This estimate was 7.5 Å regardless of the considered conductance level.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Ability to form channels in biomembranes: a structural scheme common to the whole beticolin family?

We have previously shown that B0 is able to form channels in the plasma membrane of plant protoplasts or Xenopus laevis oocytes, as well as in artificial bilayers (Goudet et al., 1998). B3 differs from B0 by an isomerization (para versus ortho) and the R residue (see Fig. 1 A), which is C₃H₄O₂ for B3 and C₃H₄ for B0. With the present observation that B3 is also able to form channels in bilayers, it might be proposed that this property is a general feature of the whole beticolin family regardless of isomerization and R-substitution type.

The free Mg²⁺ and B3 concentrations used in the current experiments were chosen based on previous B0 studies. Macroscopic currents could be recorded in the presence of high Mg²⁺ concentration (data not shown). The low used free concentration of Mg²⁺ (10 µM) made it possible to record reproducibly "single-channel" rather than macroscopic B3 currents and to observe the multiple conductance behavior of B3 channels.

X-ray diffraction of B3 microcrystals allowed the determination of the 3D structure of the toxin molecule, the length of which is ~17 Å (Fig. 1) (Prangé et al., 1997). Therefore, a multimeric assembly is likely to be necessary to form a pore across a 30-Å-thick bilayer (Aidley and Stanfield, 1996). Beticolins can assemble themselves into dimeric structures in the absence of Mg²⁺ (Prangé et al., 1997). However, in the presence of Mg²⁺, a more stable, electrically neutral dimer assembles, which is made of two beticolin molecules and two Mg²⁺ ions (Gomès et al., 1996). Interestingly, this association with Mg²⁺ strongly increases the partitioning of the toxin into the hydrophobic phase (Mikes et al., 1994). The 3D structure of the complex made of two B1 molecules and two Mg²⁺ ions was elucidated previously (Jalal et al., 1992). B1 is a para-beticolin, very similar to B3: the only difference is a CH₃ in B1 that is replaced by a CH₂OH in B3 (within the R-residue). Comparison of the 3D structures proposed for any of these molecules shows that this substitution has a poor effect on the whole backbone shape (Prangé et al., 1997). Therefore the structural scheme demonstrated by Jalal et al. (1992) for the (B1-Mg)₂ complex can be extrapolated to the (B3-Mg)₂ complex. The former complex is shaped like a slightly twisted but open rectangular (~5 × 10 Å) cage (Jalal et al., 1992). We propose, as illustrated in Fig. 6, that the (B3-Mg)₂ complex has a similar design. This (B3-Mg)₂ complex would be the basic element of the channels formed by B3. One can imagine that the stacking of two or three such elements would constitute a tubular structure long enough to cross a bilayer and therefore to allow the transmembrane movement of ions within itself. According to our current records, the first conductance level (1) may correspond to such a structure. According to data in Figs. 4 and 5, the pore radius would be 7.5 Å. This value, derived from effects of NEs, is a maximum estimate of the actual pore size. Furthermore, it is likely that the pore size is smaller for a B3 channel being part of a crystal (Fig. 6 B) than for a B3 channel freely "floating" in a lipid bilayer (data in Figs. 4 and 5). Therefore, there is no contradiction between our data and the above discussed structural model.

View larger version (30K): [in this window] [in a new window]   FIGURE 6   Structural model for the B3 elementary channel. (A) Molecular structure of a B3 dimer assembled with two Mg2+ (adapted from Jalal et al., 1992). (B) Stereo view of a B3-Mg2+ dimer deduced from the study of x-ray diffraction by B3 microcrystals (Prangé et al. 1997). This complex is a twisted rectangular box, the central lumen of which is assumed to make the pathway for ions (along a vertical axis perpendicular to the picture). The section of the lumen would be 10 × 5 Å (see scale bar), and three (or four) stacked dimers may form a pathway long enough (30 or 40 Å) to cross the bilayer.

Similar ionic selectivity and pore size of multiple conductance levels of B3 channels: probable cluster organization of B3 channels

Incorporation of B3 into planar lipid bilayers resulted in single-channel events displaying multiple amplitudes, with most of the events (~80%) occurring from the baseline (Fig. 2). This suggests that multiple current amplitudes were not due to superimposed openings of several B3 channels, but rather supports the hypothesis that B3 is able to form channels endowed with multiple conductance levels. Two hypotheses can be considered regarding the type of channel structures that could underlie such behavior: 1) a cluster organization of elementary channels that open and close simultaneously or 2) different channel structures. For example, these structural organizations have been proposed for channels formed, respectively, by syringomycin E (Feigin et al., 1996; Kaulin et al., 1998) and by alamethicin (Meves and Nagy, 1989; Woolley and Wallace, 1992; Bezrukov and Vodyanoy, 1993; Vodyanoy et al., 1993).

Regarding B3 channels, the fact that the first three conductance levels are multiples of each other is a strong support for the cluster hypothesis. As shown in Fig. 6, some free phenol groups, turned out from the structure, are present in the (B3-Mg)₂ complex and could favor the proposed clustering process, which would involve a decrease in the surface polarity of the assembly. Further support for the cluster model was gained from ionic selectivity and pore size studies.

Reversal potentials were determined in asymmetrical cis/trans solutions of KCl, NaCl, LiCl, NH₄Cl, or TEACl. The permeability ratios derived from the data, using the Goldman-Hodgkin-Katz equation, indicated that B3 channels were poorly selective, as previously reported for B0 channels (Goudet et al., 1998). This poor ionic selectivity is in agreement with the pore size value discussed above. However, in any ionic condition tested, reversal potential values were very close to each other for the first three levels of conductance (Fig. 3 and Table 1). This finding indicates that channel structures underlying the first three conductance levels have similar ionic selectivities.

It is widely accepted that the ionic selectivity of ion channels results from interactions between permeating ions and the pore wall (Imoto, 1993). Therefore, similar ionic selectivities for the different conductance levels of B3 channels suggest similar pore structure and size, as they would be within clusters. Indeed, NEs of different hydrodynamic radii similarly influenced the first three conductance levels of B3 channels (Figs. 4 and 5). Therefore, the pore size estimate that could be deduced from the data was the same for any of 1 to 3. This finding strongly supports the hypothesis of a cluster organization of B3 channels.

As discussed above, elementary B3 channels (corresponding to 1) would have a (maximum) 7.5-Å pore radius and a 16-pS single-channel conductance (in 200 mM KCl). Clusters would be made of pairs or tetrads of such elementary channels (2 being twice 1, and 3 twice 2).

The elementary conductance of B3 channels is in the same range as that of ion channels formed by toxins with similar pore size: e.g., syringomycin E (~30 pS in 0.1 M NaCl/20 Å pore diameter) (Kaulin et al. 1998), -hemolysin (~40 pS in 0.3 M KCl/~14 Å pore diameter) (Menestrina 1986; Song et al. 1996), and colicin Ia (~90 pS in 1.77 M KCl/~14 Å pore diameter) (Krasilnikov et al. 1998).

Biological significance

In the present study, B3 is found to form poorly selective ion channels. Moreover, the pore size of B3 channels (r = 7.5 Å) is such that biological solutes (e.g., glucose and sucrose, Figs. 4 and 5) are able to permeate through these channels. These findings could be consistent with previous data indicating that, in plant cells, beticolins induced a loss of electrolytes, amino acids, and betacyanine (Macri and Vianello, 1979) and depolarized transmembrane potential (Gapillout et al., 1996). Recently, we reported that another toxin of this family, beticolin 0, is also able to form poorly selective ion channels and thereby to strongly increase the membrane conductance of plant cells, animal cells, or planar lipid bilayers (Goudet et al., 1998). As channel formation by beticolin 0 seems strictly dependent on free Mg²⁺ availability, macroscopic currents can be obtained by increasing either beticolin or Mg²⁺ concentration and suppressed by, e.g., catching Mg²⁺ with an appropriate EDTA concentration (Goudet et al., 1998). Whether a low or a macroscopic conductance is induced by beticolin under physiological conditions is still unknown. Mg²⁺ concentration, which is much higher in vivo than in our conditions (10 µM), may make macroscopic beticolin conductance possible. On the other hand, it reasonably could be assumed that the beticolin concentration is much lower in vivo than in our experiments (10 µM) and therefore may limit the number of channels formed. In vivo, the channel-forming feature of beticolins may result in the collapse of ionic and electrical gradients. The ability to form channels, therefore, is likely to be involved in the biological activity of beticolins, as hypothesized for syringomycin, a necrosis-inducing lipopeptide toxin secreted by the phytopathogenic bacteria Pseudomonas syringae (Hutchison et al., 1995).