INTRODUCTION

TOP ABSTRACT INTRODUCTION TYPES OF CYTOTOXIC PEPTIDES STRUCTURE OF CYTOTOXIC PEPTIDES BIOLOGICAL ACTIVITY OF... CYTOTOXINS AS FORMERS OF... ION CHANNEL CHARACTERISTICS OF... SPECIALIZED REGIONS OF PROTEIN... CYTOTOXIC CHANNELS UNDER... PHYSIOPATHOLOGICAL SIGNIFICANCE... CONCLUSIONS REFERENCES

CYTOTOXIC PEPTIDES ARE SMALL PROTEINS that interact with lipid bilayers, resulting in an alteration of the cell's membrane permeability that leads to cell death. The peptides are thought to contribute to this process through the formation of new ion channels in the cell membrane and/or by changing the activity of existing channels. Many cytotoxic peptides have been shown to act against bacterial cells, e.g., magainin I and II (24, 31), Cyprinus carpio peptides (55), a 9-kDa polypeptide originally extracted from porcine lymphocytes (NK-lysin) (1-5), melittin (reviewed in Ref. 13), lycotoxins (87), insect defensin (20), and scorpion defensin (21). Dermaseptin acts against fungal cells and pilosulin 1 attacks mammalian cells (67, 86). Mammalian defensins are active against all enveloped viruses (43).

The characterization of cytotoxic channel-forming peptides, which modify the cellular signal transduction, forms the basis of understanding how they function in the living organism. This is of great significance to the fundamental science needed for the design of new pharmaceutical agents and for insight into cytotoxic diseases such as Creutzfeldt-Jacob disease or "mad cow" disease. Aside from the benefits flowing from improvement to health and pharmaceutical development, the investigations of cytotoxic peptides could have a marked impact on the meat industries. As shown in the United Kingdom, the meat industry needs to understand and combat cytotoxic disease, which can have devastating economic consequences, while efficiently minimizing the more common pathologies that increase production costs and reduce market acceptance.

Because the molecular mechanisms of interactions between proteins and membranes are an important frontier of cell biology, the investigations of these peptides could make a significant contribution by clarifying the means of interaction between small peptides and cell membranes and by indicating how this process can be modulated. The new information about the membrane-binding properties of peptides, which underlie intracellular signaling, is of fundamental benefit to this increasingly vital field of cytotoxic peptide research. The understanding of the properties of cytotoxic peptides could allow us to 1) formulate models of bilayer insertion and conditions of lipid composition, voltage, pH and Ca²⁺, which may be required for peptide interaction and channel formation; 2) model how highly charged, water-soluble peptides come together and interact with the membrane to form channels; 3) determine the roles of peptide domains in the biophysical and pharmacological properties of the channel functions by locating configurations, regions, and polar and nonpolar amino acids that are involved in determining the conductance levels, ionic permeability and selectivity, voltage dependency, gating, and kinetics of the channel; 4) clarify the domain structure of cytotoxic pore-forming peptides, e.g., prion and amyloid -protein (AP), an apparently disparate group of cytotoxins that may possess some common structural features; and 5) provide models of how these peptides can modify cellular physiological functions and cause pathologies.

Cytotoxic peptides have been isolated from a variety of natural sources, including venoms and antimicrobial secretions, and are involved in the mammalian immune response, being present in a range of cells of the immune system. In addition, there is some evidence that molecules of similar activity may be involved in "autocytotoxic" conditions, where the body produces peptides that act against its own cells, causing disease. Possible examples include the role of amylin in type II diabetes and apoptosis (11, 65), the prion peptide in the mad cow disease (58), and AP in Alzheimer's disease (6-9, 66, 73). It is thought that these cytotoxic-formed ion channels cause these diseases by modifying a second messenger system, e.g., Ca²⁺ homeostasis (see Fig. 7B).

The structure, biological activity, and electrical activity of some cytotoxic peptides found in toxins and antimicrobial secretions have been investigated extensively. This information can be used to characterize molecules with cytotoxic activity and may aid in the identification of other, naturally occurring, cytotoxic peptides. Interest in the identification and characterization of such molecules stems from their potential as pharmacological agents, either as antimicrobial or tumoricidal compounds. Knowledge of the mechanisms of their action also provides insight into conditions caused by similar molecules and possible treatments for such conditions. This review examines a selection of not-well-characterized cytotoxic molecules found in venoms, antimicrobial secretions, and the mammalian immune system. We have surveyed and summarized our current understanding of the properties of these cytotoxic-peptide-formed ion channels and provided a framework for future research. These peptides are grouped according to their having several common features. They are small molecules, often cationic and amphipathic in nature. Their amino acid sequences vary, but certain motifs and secondary structures are common. Few of the ion channels formed by cytotoxic peptides have been characterized extensively, but those that have show very diverse characteristics and include anion-selective, cation-selective, and nonselective channels. Some cytotoxic peptides have been well characterized and reviewed previously (see Ref. 13). The natriuretic peptides, some of which are present in venoms and toxins, have been reviewed elsewhere (46-49).

	TYPES OF CYTOTOXIC PEPTIDES

TOP ABSTRACT INTRODUCTION TYPES OF CYTOTOXIC PEPTIDES STRUCTURE OF CYTOTOXIC PEPTIDES BIOLOGICAL ACTIVITY OF... CYTOTOXINS AS FORMERS OF... ION CHANNEL CHARACTERISTICS OF... SPECIALIZED REGIONS OF PROTEIN... CYTOTOXIC CHANNELS UNDER... PHYSIOPATHOLOGICAL SIGNIFICANCE... CONCLUSIONS REFERENCES

The cytotoxic peptides discussed here can be divided into three main groups based on the origin of the molecule. Perhaps the most well-studied group is that of the cytotoxic peptides found in various venoms. Melittin, found in bee venom, is particularly well characterized, providing a model for other cytotoxic molecules (see review in Ref. 13). Cytotoxic peptides are also found in scorpion venom (long- and short-chain scorpion toxins), jumper ant (Myrmecia pilosula) venom (pilosulin 1), and the venom of the wolf spider (Lycosa carolinensis; lycotoxin I and II). The second group of cytotoxic peptides is found in antimicrobial secretions and cells involved in immunity. Magainin I and II secreted from Xenopus laevis skin are the most extensively characterized (also reviewed in Ref. 13). Dermaseptin is another peptide isolated from frog skin. Antimicrobials have been identified in skin secretions of carp [C. carpio (55)] and in cells of the immune system (e.g., insect defensins, scorpion defensins, mammalian defensins, and cryptdins). The third group of cytotoxic peptides is involved in pathological conditions and is believed to be produced by the organism itself (autocytotoxicity). Amylin, which is cytotoxic to pancreatic -cells and implicated in the pathogenesis of type II diabetes, is thought to act in this way (65). Also, fragments of the prion peptide are believed to form or alter ion channels causing cytotoxicity, and these can be considered in this group, because the prion peptide is a naturally produced peptide in mammalian systems. Last, synthetic peptides based on the sequences and properties of naturally occurring cytotoxic peptides have been demonstrated to retain cytotoxic activity. These can be distributed within the above three groups according to the cytotoxic peptide from which they are derived (e.g., see Refs. 19, 22).

	STRUCTURE OF CYTOTOXIC PEPTIDES

TOP ABSTRACT INTRODUCTION TYPES OF CYTOTOXIC PEPTIDES STRUCTURE OF CYTOTOXIC PEPTIDES BIOLOGICAL ACTIVITY OF... CYTOTOXINS AS FORMERS OF... ION CHANNEL CHARACTERISTICS OF... SPECIALIZED REGIONS OF PROTEIN... CYTOTOXIC CHANNELS UNDER... PHYSIOPATHOLOGICAL SIGNIFICANCE... CONCLUSIONS REFERENCES

Overall Characteristics

Sequence and Structure

Jones (42) made prediction of the tertiary structure of NK-lysin using multiple sequences and recognized supersecondary structural motifs. Similarly, Liepinsh et al. (57) used the NMR structure of NK-lysin to reveal folding in saposin. NK-lysin is viewed as the first representative of a family of sequence-related proteins, such as saposins, surfactant-associated protein B, pore-forming amoeba proteins, and domains of acid sphingomyelinase, acyloxyacylhydrolase, and plant aspartic proteinases, for which a four--helix bundle motif of cytolytic peptides is suggested (27).

Cationic molecules with evenly distributed positive residues. Many of the cytotoxic peptides are thought to form structures composed primarily of -helices when in a hydrophobic environment such as that encountered in the lipid membrane. They are often compared with melittin, one of the first cytotoxic molecules whose structure was determined. Vogel and Jahnig (84) used Raman spectroscopy to determine the three-dimensional shape of melittin in lipid membranes, finding that the molecule formed a bent -helix from residues 1 to 21. The helix is divided into residues 1-11 and 12-21 that are bent at an angle of 60° to each other. The COOH-terminal residues, 22-26, form a nonhelical hydrophilic domain Fig. 1 (see Ref. 84). The venom of the wolf spider, L. carolinensis, contains two peptide toxins: lycotoxin I and II. The amino acid sequences of lycotoxin I (25 amino acids) and lycotoxin II (27 amino acids) are characterized by lysine residues every four to five amino acids. Similar motifs are found in adenoregulin, dermaseptins, and magainins (as determined by a GenBank search). Both lycotoxin molecules are positively charged (lycotoxin I = +5; lycotoxin II = +6). The predicted secondary structure of these two toxins is that of an amphipathic -helix (87).

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Schematic model of the conformation of water-dissolved and membrane-bound melittin. Two -helical segments of residues 1-11 and 2-21 are bent by 60° relative to each other, the COOH-terminal segment of residues 22-26 being nonhelical. The convex side of the bent helix is hydrophilic, the concave side hydrophobic. Circles denote positively charged residues. [From Vogel and Jahnig (84), reprinted with permission from the Biophysical Society.]

Cysteine-rich molecules composed of both -helix and -sheet domains. Scorpion toxins, scorpion defensins, and insect defensins form a second, structurally similar, group of cytotoxic molecules. Another related peptide group are the thionins (50). The molecules have poor sequence homology but are all cysteine rich and share a similar cysteine motif. These cysteine molecules form disulfide bridges that stabilize a common secondary structure composed of an antiparallel triple-stranded -sheet linked by two disulfide bridges to an -helix; the -sheet is linked via a third disulfide bridge to the NH₂-terminal region of the molecule (Fig. 2; see also Refs. 15-16, 50). This structure forms the core of the molecules in this group. There have been three groups of scorpion toxins identified.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Schematic representation of secondary structures and disulfide bridges of toxin III from the scorpion L. quinquestriatus quinquestriatus (Lqq III), 1-purothionin from wheat (1-P), and defensin A from P. terranovae (Def A). Helices and extended strands are depicted as barrels and arrows, respectively. [From Landon et al. (50), reprinted with permission from Eur J Biochem.]

View larger version (37K): [in this window] [in a new window]   Fig. 3.   Stereoviews of a space-filled model of toxin III. Hydrophobic residues are in yellow, aromatic residues in purple, basic residues in cyan, acidic residues in red and other residues in green. A and B: views that differ by a 180° rotation around the vertical axis. A stereoview of the backbone is given above each space-filled model. [From Landon et al. (50), reprinted with permission from Eur J Biochem.]

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Alignment of the sequence of the scorpion defensin with the sequences of insect defensins (20). The defensins were characterized from species belonging to three insect orders: Diptera (a), Coleoptera (b), and Odonata (c). The scorpion defensin is closely related to the Aeschna defensin (c). Dashes indicate gaps to optimize the alignment. Identical amino acids are boxed. Boxes in bold represent the identical residues between Aeschna and scorpion residues. [From Cociancich et al. (21), reprinted with permission from Academic Press, Inc.]

Molecules composed predominantly of -sheet structures. Yet another group of peptides with cytotoxic activity are those composed predominantly of -sheet structures. These include mammalian defensins and cryptdins, human amylin, and PrP-(106-126). Of particular interest are those that were predicted on the basis of their amino acid sequence to form -helices but that have been demonstrated to be composed of -sheet when active. This raises the question of whether other peptides predicted to form -helices in fact do so. Investigations of ion channel-forming molecules in other areas has revealed that those forming -sheets represent an important structural class of ion channel-forming molecules (6-9, 73).

Unknown sequence and/or structure. The amino acid sequence and/or secondary structure of a number of cytotoxic peptides has not been determined. Investigation of the structures of such molecules is needed if we are to gain a more comprehensive picture of the patterns of channel formation by cytotoxic peptides. Two peptides with antibacterial activity were isolated from the skin secretions of C. carpio (55). The 27-kDa protein isolated was found to be glycosylated, and its first 19 amino acid residues were determined. It did not show any similarity to known protein sequences. The 31-kDa peptide was nonglycosylated and able to form ion channels. It could not be sequenced because it was blocked at its NH₂-terminus. NK-lysin, isolated from pig small intestines and believed to originate from cytotoxic T cells and NK cells, is 78 amino acid residues in length. It contains six cysteine residues, which form three disulfide bridges (3, 5, 17, 54). The molecule is much larger than the mammalian defensins and, despite the disulfide bridges, shows no sequence similarity to these molecules (3, 5).

Tertiary Structure: Models of Channel Formation

View larger version (82K): [in this window] [in a new window]   Fig. 5.   Schematic model of tetrameric melittin in membranes. Spheres, nonhelical COOH-terminal segments of melittin; cylinders, membrane-bound -helical segments with the hydrophobic sides (white) facing the bilayer lipids and the hydrophilic sides (gray) facing each other, thus forming a bilayer-spanning polar core. Tryptophan residues (small open circles) are shown to illustrate the position of fluorescence donors and acceptors. [From Vogel and Jahnig (84), reprinted with permission from the Biophysical Society.]

Cruciani et al. (24) proposed three structural models for a magainin II-formed ion channel (Fig. 6). In the models, the magainin II molecules are arranged in dimers of -helices aligned to form antiparallel amphipathic units. Type 1 channels were composed solely of peptide molecules arranged so that the polar side chains of amino acids extended into the lumen, while the hydrophobic portions of the molecules formed the wall of the channel. Large numbers of monomers were required to form channels of this type. The central pore of the channel was calculated to be electropositive, making the channel selective for anions. These channels are similar to those described for melittin above. Type 2 channels were composed of both peptides and lipids, with some of the channel wall being formed by the hydrophilic heads of lipid molecules from the surrounding lipid bilayer. The alkyl tails of lipid molecules involved in channel formation extended into the hydrophobic region of the membrane and were stabilized by aggregates of magainin II peptide on the surface of the membrane, where hydrophobic lipid tails would otherwise come into contact with the aqueous environment. The central pore of the channel thus formed was calculated to be electronegative, consistent with cationic selectivity. Type 3 channels also involved both lipid and peptide molecules. In this case head groups from lipid molecules formed the lining of the ion channel while the peptide molecules remained on the surfaces of the membrane stabilizing the structure. Magainin II is required on both sides of the membrane for this type of channel to form. In this model the channel lumen was calculated to be electronegative, consistent with cationic selectivity. The channel models were dependent on the lipid composition of the membrane in which the channel was formed. Cruciani et al. (24) based their models on membranes composed of a combination of negatively and positively charged lipids. Type 2 structural models assumed that 24 out of the 27 lipid molecules involved in forming the pore were palmitoyl-oleoyl-phosphatidylserine (PS), while type 3 models were constructed using palmitoyl-oleoyl-phosphatidylethanolamine (PE) to line the channel and PS to form the bilayer regions.

View larger version (44K): [in this window] [in a new window]   Fig. 6.   Examples of the channel models formed as viewed looking down the channel axis. A: type I. The channel is formed by 12 antiparallel magainin -helices. Hydrophilic side chains extend inward into the pore, and hydrophobic side chains extend outward into the lipid. The magainin helices are displayed as -carbon traces with the NH2- and COOH-terminal amine and carboxylate groups included. The lysine and phenylalanine residues are also included to indicate the polarity of the helical faces. B: type II. A combination of magainin 2 antiparallel dimers on the transmembrane bound to the lipid head groups (a) and membrane surface (b). The lipid molecules are displayed as circles for the head groups and straight lines for the alkyl chains, parallel to the membrane plane. C: type III. All of the magainin 2 dimers on the lipid head groups completely forming the channel lining (a) and the surface (b). The type III model is illustrated as part of a hexagonal lattice. Squiggly lines represent alkyl chains perpendicular to the membrane plane. [Reprinted from Eur J Pharmacol, vol. 226, Cruciani RA, Barker JL, Durell SR, Raghunathan G, Guy HR, Zasloff M, and Stanley EF. Magainin 2, a natural antibiotic from frog skin, forms ion channels in lipid bilayer membranes, p. 287-296, copyright 1992, with permission from Elsevier Science (24).]

Cruciani et al. (24) were then able to speculate on the type of channel that was formed by magainin II in their experiments. Because the channels that they found were cationic selective, type 1 channels were not probable. They speculated that type 3 channels accounted for the majority of conductance. However, because of the requirement for peptide on both sides of the membrane for the formation of type 3 channels, these were not the initial channels formed. They proposed that type 2 channels were formed first and then, as they broke down, released some peptide on the other side of the membrane, allowing type 3 channels to form. Extending their work to other literature, Cruciani et al. (24) speculated that the magainin I anionic selective channels, found by Duclohier et al. (31), may have been type 1 channels. However, this work was done using neutrally charged lipid membranes, and these models of channel formation were formed on the basis of some negatively charged lipids in the membrane. They acknowledged that determining the likely channel model was not straightforward. More complex models involving lipids from the membrane-forming sections of the channel have been suggested for the formation of some channels, in particular to explain how cationic peptides could form a cationic-selective channel. Models of channel formation for cytotoxic peptides not composed solely of amphipathic -helices have not been investigated in such detail. The family of scorpion toxins and scorpion defensin toxins are proposed to interact with preexisting ion channels in the membrane. However, the related peptide, insect defensin, itself forms ion channels (20), suggesting that peptides from this structural group can form ion channels. Further investigation will help to clarify this.

	BIOLOGICAL ACTIVITY OF CYTOTOXIC PEPTIDES: EVIDENCE FOR CYTOTOXIC ACTIVITY

TOP ABSTRACT INTRODUCTION TYPES OF CYTOTOXIC PEPTIDES STRUCTURE OF CYTOTOXIC PEPTIDES BIOLOGICAL ACTIVITY OF... CYTOTOXINS AS FORMERS OF... ION CHANNEL CHARACTERISTICS OF... SPECIALIZED REGIONS OF PROTEIN... CYTOTOXIC CHANNELS UNDER... PHYSIOPATHOLOGICAL SIGNIFICANCE... CONCLUSIONS REFERENCES

By definition cytotoxic peptides have cytotoxic activity, mostly antimicrobial activity or hemolytic activity. This activity is demonstrated in assays of cytotoxicity against various cell types (bacterial, fungal, and mammalian cells, especially blood cells). The specific biological activity of each peptide depends on the environment in which it is found. Some peptides found in venom function as toxins and are used by organisms for defense or to subdue prey. Cytotoxic peptides found in antimicrobial secretions function primarily by providing defense against pathogens such as colonizing bacteria. Those produced by immune system cells are also involved in host defense. There is some evidence that molecules with a similar molecular activity are also involved in autocytotoxic conditions.

Lycotoxins I and II, from wolf spider venom, were shown to have antimicrobial activity against both prokaryotic and eukaryotic cells. Various microbes were incubated with the peptides to determine the specificity of their antimicrobial activity. Lycotoxin I was inhibitory at concentrations as low as 5 µM for one gram-positive species (Bacillus thuringiensis israelensis), whereas lycotoxin II was most active against gram-negative strains (Escherichia coli strain DH5 minimal inhibitory concentration 40 µM). Both peptides were active against yeast (Candida albicans) at a concentration of 40 µM. Magainin II was also assayed to provide comparison: in most instances the inhibitory activity exhibited by the lycotoxins was stronger than that exhibited by magainin II (87). Inhibition of growth occurred at a critical concentration, indicating self-aggregation of the peptide. Hemolytic activity was also assayed for lycotoxin I. Lycotoxin I was found to cause lysis of erythrocytes at concentrations >100 µM and was more active than magainin II in this respect (87). Pilosulin 1 causes cell lysis via disruption of membrane integrity in both dividing and nondividing blood cells (86). Lysis of erythrocytes was complete at a concentration of 40 µM, with partial lysis occurring at concentrations as low as 1.25 µM. White blood cells were differentially affected by pilosulin 1, with mononuclear cells being more susceptible to lysis than granulocytes. Lysis was rapid and usually complete, but results (for white blood cells) differed in normal individuals by up to fivefold. Pilosulin 1 also had a cytotoxic effect on Epstein-Barr virus (EBV)-transformed B lymphocytes [dose of toxin that will kill 50% of test subjects in standard time (LD₅₀) = 0.4 µM] (86).

Lemaitre et al. (55) incubated various strains of bacteria with two ion-channel-forming peptides (27 and 31 kDa) isolated from carp (C. carpio) skin mucous. Both compounds were found to have antibacterial activity. Inhibition of bacterial growth occurred at concentrations of ~5 µg/ml (0.16-18 µM). Gram-positive and gram-negative bacterial strains were equally affected. However, Micrococcus luteus (gram positive) showed great sensitivity when exposed to the 27-kDa peptide at a concentration of only 0.5 µg/ml (0.018 µM) (55).

NK-lysin was shown to have antimicrobial activity against several strains of gram-negative and gram-positive bacteria, the highest activity being against E. coli and Bacillus megaterium. In assays for hemolytic activity against sheep red blood cells, no hemolysis was detected at concentrations of NK-lysin of 170 µM. The peptide also exhibited antitumor activity in assays against YAC-1 tumor cells (lymphoma cells from the lung) (4). In addition, it has been shown that NK-lysin (1-100 nM) potently and reversibly stimulates insulin secretion in rat pancreatic islets and in the  cell line HIT T15 (89). However, the effect of NK-lysin was not accompanied by changes in Ca²⁺ concentration. Furthermore, the stimulatory activity of NK-lysin on insulin release was observed in permeabilized islets under Ca²⁺-clamped conditions. These findings indicate that NK-lysin action, although it may interact with the membrane, is unlikely to be due to NK-lysin-formed Ca²⁺-permeable channels.

RK-1, a novel agent related to the corticostatins/defensins and isolated from the kidney, was found to have antimicrobial activity. It was active against E. coli cultures at concentrations between 15 and 150 µg/ml (12). Magainin I and II show antibacterial activity against a range of microbes, including both gram-positive and gram-negative bacteria as well as fungi (90). Magainin II has between 5 and 10 times the antimicrobial activity of magainin I and is more abundant in the secretions of X. laevis. Magainins also have anti-protozoan activity, with lysis of the protozoan Paramecium caudatum occurring on exposure to magainin II at concentrations of 10 µg/ml (90).

Selected magainin analogs with amino acid substitutions aimed at increasing the amphipathic -helical nature of the molecule have been shown to have increased antimicrobial activity (19). The synthetic peptides had the same spectrum of activity as the natural peptide but were more active by one to two orders of magnitude. Whereas magainin I does not cause hemolysis of erythrocytes at concentrations of 250 µg/ml, the altered analogs (except for analog H) were capable of causing hemolysis at concentrations of 100 µg/ml. Dermaseptin was tested for cytotoxic activity against fungal species (Aspergillus fumigatus and Arthroderma simii) known to be pathogenic to the frog P. sauvagii. The fungal species were inhibited at a dermaseptin concentration of 10 µg/ml. Assays against a gram-positive bacterial species (Bacillus subtilis) showed no inhibition (67). The synthetic peptides that were constructed by Cornut et al. (22) were assayed for hemolytic activity with the use of human red blood cells. Hemolytic activity varied according to the length of the peptide, with the longer peptides being more active. Activity was recorded even at very low concentrations (LD₅₀ = 4-6 × 10⁸ µM). This corresponded to an activity 6-10 times higher than that found for melittin (22). On the basis of this cytotoxic activity, such peptides are suspected to cause alterations in membrane permeability. Generally, it appears that some peptides do this via the formation of ion channels, whereas others may interact with ion channels already present in the membrane.

	CYTOTOXINS AS FORMERS OF ION TRANSPORT PATHWAYS: EVIDENCE FOR ION CHANNEL FORMATION

TOP ABSTRACT INTRODUCTION TYPES OF CYTOTOXIC PEPTIDES STRUCTURE OF CYTOTOXIC PEPTIDES BIOLOGICAL ACTIVITY OF... CYTOTOXINS AS FORMERS OF... ION CHANNEL CHARACTERISTICS OF... SPECIALIZED REGIONS OF PROTEIN... CYTOTOXIC CHANNELS UNDER... PHYSIOPATHOLOGICAL SIGNIFICANCE... CONCLUSIONS REFERENCES

Ion Flux Experiments

Lycotoxin I was found to cause an efflux of Ca²⁺ from rat brain synaptosomes and to dissipate the membrane potentials of insect muscle cells (87). Addition of lycotoxin I to Ca²⁺-loaded rat synaptosomes resulted in dissipation of the Ca²⁺ gradient, mediated through an efflux of Ca²⁺ from the synaptosome, and the prevention of Ca²⁺ sequestration. Similarly, the membrane potentials of insect muscle cells were dissipated when the cells were exposed to lycotoxin I at a concentration of 3 µM. In combination with the antimicrobial data, these factors constitute indirect evidence that lycotoxins act as pore-forming peptides (87). However, detailed characterization of these potential pores, including ion channel kinetics, selectivity, and conductance, was not completed.

Florio et al. (36) report that the PrP-(106-126) fragment induces apoptosis in GH3 cells via dose-dependent inactivation of the L-type voltage-sensitive Ca²⁺ channels. The blockage of these channels resulted in blockade of the increase in cytosolic Ca²⁺ following depolarization and resulted in cell death with features of apoptosis. They speculated that alterations in Ca²⁺ homeostasis may be the trigger for apoptosis. In contrast, Lin et al. (58) found that PrP-(106-126) in planar lipid bilayers induced the formation of ion channels that were permeable to common physiological ions. It was proposed that the channels would cause leakage of ions across the membrane of cells, resulting in disturbance of ion balance. This could lead to a large metabolic demand being placed on the cells with attempts to correct the ion balance. Disturbance of ion balance could also lead to ionic toxicity and might trigger apoptosis. Discharge of membrane potential, due to ion imbalance, could alter cellular Ca²⁺ levels (via voltage-dependent Ca²⁺ channels, N-methyl-D-aspartate receptors, or alteration in Na⁺/Ca²⁺ exchange mechanisms), which may trigger apoptosis (58, 69).

More recently, Lin et al. (58) used a combined immunofluorescence labeling with an antibody raised against the NH₂-terminal domain, and atomic force microscopy, to image the structure of AP-reconstituted phospholipid vesicles, whose ⁴⁵Ca²⁺ uptake was measured to assess the Ca²⁺ permeability across the vesicular membrane. Their findings suggest that globular and fresh AP-(1-40) forms Ca²⁺-permeable channels. For example, ⁴⁵Ca²⁺ uptake was inhibited by a monoclonal antibody raised against the NH₂-terminal region of AP and by Zn²⁺, a known blocker of AP-formed channels (58).

Villus Enterocyte Volume Assay

MacLeod et al. (60) demonstrated the effects of the family of corticostatic peptides and defensins on villus enterocytes. The peptides, when incubated in isosmotic conditions with enterocytes from guinea pig jejunums, caused a reduction in cell volume. They suggested that the reduction in volume was due to activation of L-type Ca²⁺ channels (60). By using the villus enterocyte volume assay, it was shown that RK-1, like the corticostatin/defensins, causes enterocyte shrinkage of epithelial cells in isotonic media. This shrinkage occurred in both the presence and absence of Na⁺ and was prevented by the use of Ca²⁺-free media. Niguldipine prevented the volume reduction caused by RK-1. Activation of dihydropyridine-sensitive Ca²⁺ channels is implicated in the shrinkage of the cells (12).

The cryptdins, or enteric defensins, are secreted by cells lining the mammalian gut and have been found to form anion-conductive channels in human intestinal T84 cells in vitro (56). Use of the villus enterocyte volume assay to assess ion movement across the cell membrane showed that the cryptdins elicited a Cl secretory response. The secretory response was not attributable to receptors or ion channels already present in the cells and was correlated with the formation of new anion-permeable channels in the cell membrane. The pores formed were permeable to carboxyfluorescein.

Electrical Properties

Most experiments to characterize ion channels formed by cytotoxic peptides have used the patch-clamp or lipid bilayer techniques (Table 1). Ion channel experiments directly illustrate the capacity of peptides to form ion-permeable channels. Relatively few peptides known to have cytotoxic activity have been proven to form ion channels in such experiments. Those that do form ion channels include insect defensin (20), melittin (71, 72), C. carpio peptides (55), the magainins (24, 31), mammalian defensin rabbit NP-1 (43), AP-(1-40) (9), amylin (65), and PrP-(106-126) (58). Some of these channels have also been further characterized to determine their conductance, current-voltage relationships, and ion selectivities.

The characteristics of ion channels formed by cytotoxic peptides vary considerably. For example, magainin I forms anion-selective channels (31) and magainin II forms cation-selective channels (24). Rabbit NP-1, a mammalian defensin, forms voltage-dependent, weakly anion-selective channels (43). AP formed cation-selective channels (6, 9), and human amylin was also found to form voltage-dependent channels that are permeable to Na⁺, K⁺, Ca²⁺, and Cl (65). Prion peptide segment PrP-(106-126), which was reported to block L-type voltage-sensitive Ca²⁺ channels (36), was found to form nonselective ion channels (58). Insect defensin was shown to form cation-selective voltage-dependent channels in giant liposomes (20), and C. carpio peptides form ion channels of differing selectivities (55). The details of these investigations and the characterization of the ion channels appear below and in Tables 1 and 2.

View this table: [in this window] [in a new window]   Table 2.   Summary of AP-formed ion channels

	ION CHANNEL CHARACTERISTICS OF CYTOTOXIC PEPTIDES

TOP ABSTRACT INTRODUCTION TYPES OF CYTOTOXIC PEPTIDES STRUCTURE OF CYTOTOXIC PEPTIDES BIOLOGICAL ACTIVITY OF... CYTOTOXINS AS FORMERS OF... ION CHANNEL CHARACTERISTICS OF... SPECIALIZED REGIONS OF PROTEIN... CYTOTOXIC CHANNELS UNDER... PHYSIOPATHOLOGICAL SIGNIFICANCE... CONCLUSIONS REFERENCES

Conditions Influencing the Interaction of Cytotoxic Peptides With Membranes

The conductance properties and the experimental conditions for the cytotoxic peptide-formed channels are shown in Table 1. The phospholipid composition (Table 1) of bilayers used to investigate ion channel formation by cytotoxic peptides can be important. Although some peptides (e.g., NP-1) have been shown to form ion channels in membranes composed of a wide range of lipids and to have no specific lipid requirements, many peptides require a certain content of negatively charged lipids in the membrane. In some instances the characteristics of the channels formed in membranes of different composition are different (e.g., magainin I).

Kagan et al. (43) used membranes composed of PE, phosphatidylcholine (PC), and PS in a ratio of 2:2:1 (wt/wt/wt) when testing the ability of NP-1 to form ion channels. In addition to forming ion channels in this type of membrane, they found that NP-1 was able to form ion channels in membranes composed of different mixtures of these lipids, indicating that membrane composition did not affect activity (43).

For ion channels formed by magainin I, the channel amplitude was altered by membrane composition. Membranes composed of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC)/1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoserine (7:3) favored conductance levels of 80 pS, considerably lower than the conductance levels determined in membranes composed of POPC/1,2-dioleoyl-3-phosphatidylethanolamine (7:3; 366 and 683 pS) (31). Magainin II ion channel formation requires the presence of negatively charged lipids in the membrane (24). Ion channel formation, as measured by increased membrane conductance, was demonstrated in membranes composed of PS/PC or PS/PE, but not in those containing only PE or PE/PC combinations. Cruciani et al. (24) speculated that the type of lipids present in the membrane helped determine the ion selectivity of the magainin II ion channel. Thus when magainin II interacts with membranes containing negatively charged lipids, ion channels selective for cations are generated. Conversely, membranes that do not contain negatively charged lipids are unable to support ion channel formation. Ion channel formation by magainin I was similarly dependent on the lipid composition of the membrane: membranes composed of PE/PS supported channel formation, whereas those composed of PE alone or PE/PC did not induce channel formation, indicating a requirement for negatively charged lipids in target membranes. Membrane composition altered the channel-forming activity of amylin. Membranes that had a high net negative surface charge induced the highest channel-forming activity. Membranes that contained 40% negatively charged phospholipids were approximately six times more sensitive to amylin channel formation than those containing only 20% negatively charged amino acids (65). Increasing membrane rigidity by including cholesterol causes a reduction in the membrane's sensitivity to amylin.

In some investigations, the reversibility of ion channel formation has been tested. The association of amylin with the membrane appears to be irreversible: extensive washing of the side of the membrane that contained the amylin failed to reduce or eliminate the current (65). Similarly, washing of the compartment containing the prion peptide fragment did not affect conductance, indicating that the channels formed were irreversibly associated with the membrane (58).

For several peptides, channel formation occurs only when the membrane is held at a negative voltage. This reflects the state of the membrane as it would "appear" to such peptides in vivo. For example, ion channels formed from magainin I appeared only when the membrane was held at a negative voltage (31). Similarly, Kagan et al. (43) found that the steady-state conductance could be induced after negative voltages (70 to 90 mV) were applied to the membrane for 15-30 min, and increased exponentially with increasingly negative voltage. The channels formed by AP-(25-35) are voltage dependent, opening when the opposite side of the membrane is made negative with respect to the AP-(25-35)-containing side and closing when the polarity of the voltage is reversed (66). This is taken to indicate that AP-(25-35) channels would open at the resting membrane potential of neurons if the peptides were located in an extracellular or endosomal/lysosomal compartment.

The variations in conductance and kinetic properties of cytotoxic peptide-formed channels, e.g., AP-formed channels (6, 9, 40) reflect the differences in 1) peptide properties, e.g., isoforms, the ratio of -helices to -sheets, homogeneity, and aggregation; 2) bilayer properties, e.g., phospholipids' composition, presence of solvent, bilayer capacitance, and stability; 3) experimental solutions, e.g., ionic composition, ionic concentrations and gradients, buffers, pH, and Ca²⁺ levels; and 4) recording conditions, e.g., voltage, gain, recording duration, and sampling rates.

Conductance and Current-Voltage Relations

Kagan et al. (43) found that single-channel conductance values were heterogeneous, ranging between 10 and 1,000 pS. Conductance increased with NP-1 concentration, suggesting that a multimer of NP-1 may form the ion channel. Conductance also consistently increased with time, which may be due to the formation of larger channels with an increased number of the NP-1 molecules per channel. Magainin I ion channels were shown to exhibit two possible conductance levels (366 and 683 pS). Each occurred at approximately equal rates but tended not to occur in the same experiment (31). Conductance was dependent on peptide concentration, with increased macroscopic conductance occurring with increased peptide concentration. Greater concentrations of peptide resulted in increased current amplitude. This is very likely to be due to the recruitment of more subunits of the channel-forming peptide. The current-voltage relationships for channels of both conductance levels were nonsaturable and ohmic (31). Cruciani et al. (24) found that the ion channels formed by magainin II gave heterogeneous conductance values (1-300 pS). They suggested that this wide variation was due to variable channel structure occurring with dynamic incorporation and reaggregation of molecules within the membrane. However, the unitary conductance of the single-channel conductance is reported to be between 1 and 2 pS. The current-voltage relationship for channels formed by magainin II showed rectification at both negative and positive membrane voltages. This was more marked at positive voltages, with greater current flow recorded at positive voltages than at negative voltages of the same magnitude (24).

To provide evidence that ion transport that occurred in the presence of insect defensin from P. terranovae was due to the formation of ion channels by the peptide, Cociancich et al. (20) demonstrated that insect defensin formed ion channels in giant liposomes. Peptide samples were incorporated into giant liposomes composed of asolectin, and patch-clamp techniques were used to determine the single-channel conductance of ion channels thus formed. Insect defensin was found to form channels of heterogeneous conductance values (ranging from 100 to 200 pS). The