Cupiennin 1, a New Family of Highly Basic Antimicrobial Peptides in the Venom of the Spider Cupiennius salei(Ctenidae)*

A new family of antimicrobial peptides was isolated from the venom of Cupiennius salei. The peptides were purified to homogeneity, and the sequence of cupiennin 1a was determined by Edman degradation: GFGALFKFLAKKVAKTVAKQAAKQGAKYVVNKQME-NH2. The amino acid sequences of cupiennin 1b, c, and d were obtained by a combination of sequence analysis and mass spectrometric measurements of comparative tryptic peptide mapping. All peptides consist of 35 amino acid residues and are characterized by a more hydrophobic N-terminal chain region and a C terminus composed preferentially of polar and charged residues. The total charge of all cupiennins calculated under physiological conditions is +8, and their C terminus, formed by a glutamic acid residue, is amidated. Conformational studies of the peptides revealed a high helix forming potential. Antimicrobial assays on bacteria with cupiennin 1a, 1d, and synthesized cupiennins 1a* and 1d* showed minimal inhibitory concentrations for bacteria in the submicromolar range. Their lytic effect on human red blood cells was lower by a factor of 8 to 14 than the highly hemolytic melittin. Cupiennin 1a, 1b, 1d, 1a*, and 1d* showed pronounced insecticidal activity. The immediate biological effects and the structural properties of the isolated cupiennins indicate a membrane-destroying mode of action on prokaryotic as well as eukaryotic cells.

Antimicrobial peptides are ubiquitous in nature as a part of the innate immune system and host defense mechanism. They are produced by various species, both in prokaryotic and eukaryotic cells (for reviews, see Refs. [1][2][3][4][5]. Many of these peptides act within minutes by a cell-lytic/ionophoric, nonstereoselective mechanism against a broad spectrum of bacteria (6), protozoa (7,8), filamentous fungi (9, 10), tumor cells (11), and enveloped viruses (12). Additional mechanisms to kill bacteria by acting stereospecifically on molecular targets have been described (13)(14)(15). 1 As a selective response to microbial invasion, several anti-microbial peptides have been identified in the hemolymph (e.g. hemocytes) of insects (16), spiders (17), and scorpions (18). Only a few antimicrobial peptides are exclusively and constitutively present in the venom glands of insects, namely melittin (19) and crabolin (20), and in scorpions, hadrurin (21) and scorpine (22). In 1989 the first report on bactericidal peptides in spider venom (Lycosa singoriensis) was published (23); later, lycotoxins were isolated from Lycosa carolinensis (24). Spiders are hunting predators and use paralytic venoms to immobilize their prey. Most components in their venoms act on the nervous system and are enzymatically active causing cell membrane disruption and tissue necrosis. The venom of Cupiennius salei, a hunting spider found in Central America, yielded both neurotoxically active peptides (named CSTX-1 to CSTX-13) (25) and bactericidal peptides (26). To differentiate between the two groups of peptides, we named the family of antimicrobial compounds cupiennins.
Here, the purification of the cupiennin family 1 composed of four very similar peptides, named cupiennin 1a, 1b, 1c and 1d, is reported. Cupiennin 1a is identical with peptide 8, and cupiennin 1d is identical with peptide 7 in Ref. 26. Sequence analysis of the highly cationic peptides and helix projection revealed a unique structure distinctly different from that of other potentially helical cationic peptides isolated so far. The highly antimicrobial, hemolytic, and insecticidal activity and the structural properties indicate a membrane-disturbing function of the cupiennins on prokaryotic as well eukaryotic cells. These peptides may provide a powerful tool in analyzing the structural prerequisites of antimicrobial selectivity and general cytolytic membrane action. In the light of the massive increase of multi-drug-resistant bacterial infections, further structural investigations of cupiennins may prove helpful in the development of new antimicrobial peptide pharmaceutics.
Isolation of Toxins-C. salei (Ctenidae) spider maintenance, venom collection by an electrical milking procedure, separation of venom by gel filtration, cationic exchange chromatography, and reversed-phase HPLC were performed as described previously (25). Briefly, 450 l of crude venom was fractionated into nine 50-l aliquots and diluted with 150 l of 200 mM ammonium acetate buffer, pH 5.5 (buffer A). The diluted venom was separated on a Superdex 75 HR 10/30 column (Amersham Biosciences. Inc.) in buffer A, and fractions were collected as noted on the chromatogram (see Fig. 1A). Further separation of the pooled fractions was achieved by cationic exchange on a Mono S HR 10/10 column (Amersham Biosciences. Inc.) in buffer A. Elution was done with a salt gradient (2 M NaCl in buffer A, pH 5.5) as shown in Fig.  1B. Similar fractions from four chromatographies were combined. The pooled fractions were further desalted and separated by RP-HPLC on a nucleosil 300-5 C 4 column (4.6 ϫ 250 mm, Macherey & Nagel) using 100% solvent A with a flow rate of 0.5 ml/min for 0 -15 min followed by a first 10-min gradient of 1% solvent B in A/min and a second 120-min gradient of 0.4% solvent B in A/min (solvent A, 0.1% trifluoroacetic acid in water; solvent B, 0.1% trifluoroacetic acid in acetonitrile (not shown)). After each step of purification the fractions were tested for their antimicrobial activity by a plate growth assay as described previously (26). Further purification was achieved by RP-HPLC on a nucleosil 120-5 C 18 column (4 ϫ 250 mm, Macherey & Nagel) equilibrated with 22% solvent B in solvent A and a flow rate of 0.5 ml/min. Ten minutes after injection of the sample the first gradient (22-40% solvent B) was started for 115 min followed by a second 10-min gradient (40 -100% solvent B). The peptide fractions obtained were pooled (Fig. 1C) and purified in a final RP-HPLC step on a nucleosil 100-5 C 8 HD column (4 ϫ 250 mm, Macherey & Nagel) with constant 36% solvent B and a flow rate of 0.5 ml/min (Fig. 1D). This step was repeated three to five times to obtain homogeneous peptides. The purity of these peptides was confirmed by ESI-MS and amino acid analyses.
Enzymatic Cleavages-10 g of cupiennin 1a were incubated with 1 g of chymotrypsin (sequencing grade, Roche Diagnostics) in 10 mM Tris-HCl, 1 mM CaCl 2 , pH 7.5, for 4 h at 25°C. Separation of the chymotryptic peptides was achieved with RP-HPLC on a 120 -5 C 18 column (2 ϫ 125 mm, Macherey & Nagel) equilibrated with 100% solvent A. Ten minutes after injection of the sample, a 250-min gradient of 0.2% solvent B in solvent A/min was started. Comparative digestions of 5 g of purified cupiennin 1a, 1a*, 1b, 1c, and 1d with 0.5 g of trypsin (sequencing grade, Roche Diagnostics) were carried out in 100 mM ammonium hydrogen carbonate buffer at pH 8.0 for 4 h at 25°C. Separation of the tryptic peptides was obtained under isocratic conditions with RP-HPLC on a 120 -5 C 18 column (2 ϫ 125 mm, Macherey & Nagel) equilibrated with 100% solvent A. A second run was started and after 10 min followed by a 120-min gradient of 0.23% solvent B in A/min. Solvent A: 0.1% trifluoroacetic acid in water, solvent B: 0.1% trifluoroacetic acid in acetonitrile (Fig. 3, A and B). Mass analysis of the digested samples and of the isolated peptides was performed on ESI-MS. Amino Acid Analysis-Samples were hydrolyzed in the gas phase with 6 M hydrochloric acid containing 0.1% (by volume) phenol for 24 h at 115°C under N 2 vacuum (27). The liberated amino acids were labeled with phenylisothiocyanate and the resulting phenylthiocarbamoyl amino acids analyzed by RP-HPLC on a Nova Pak ODS column (3.9 ϫ 150 mm, 4 m; Waters) in a Hewlett Packard liquid chromatograph 1090 with an automatic injection system (28).
Amino Acid Sequence Analysis-N-terminal sequence analysis was carried out either in a Procise cLC 492 protein sequencer or in a pulsed-liquid-phase sequencer 477A, both from Applied Biosystems. The released amino acids were analyzed on-line by RP-HPLC according to instructions from Applied Biosystems.
ESI-MS-The molecular mass of isolated cupiennins and proteolytic peptides was determined using electrospray ionization mass spectrometry on a single-stage quadrupole instrument (VG Platform, Micromass, Manchester, UK) calibrated with horse myoglobin in a mass range of 600 -2000 m/z. Peptides from enzymatic cleavages represent monoisotopic masses, and native cupiennins represent average masses.
Circular Dichroism-Stock solutions of the peptides were prepared by dissolving the samples in 5 mM sodium phosphate buffer at pH 7.25 containing 150 mM NaF. For CD measurements, aliquots of the solution were diluted with buffer or mixed with TFE to give a final concentration of 50 M and the desired solvent composition. Measurements were carried out on a J-720 spectrometer in 0.1-cm cells between 195 and 260 nm at room temperature (Jasco). Each spectrum was the average of five scans. The base line was subtracted. The helicity (␣) of cupiennin 1a, 1a*, and 1d* was determined from the mean residue ellipticity at 222 nm (⌰ 222 ), according to the equation: ␣ (%) ϭ (⌰ 222 ϩ 2340) ϫ 100%/ Ϫ30300 (29).
Antimicrobial Assays-Bacteria (Escherichia coli ATCC 25922; Staphylococcus aureus ATCC 29213; Enterococcus faecalis ATCC 29212, Pseudomonas aeruginosa ATCC 27853) were cultured in Mueller-Hinton broth. Determination of the minimal inhibitory concentration for the cationic antimicrobial peptides was performed using a 2-fold microtiter broth dilution assay (30). Mueller-Hinton broth was used to dilute the bacterial inoculum, which was prepared from mid-log phase cultures to give a final concentration of 1.7-3.8 ϫ 10 5 colony-forming units/ml in the wells. First, 100 l of the bacteria dilution was added to the wells followed by 10 l of the test peptides in 0.01% acetic acid, 0. Hemolytic Assay-The hemolytic activity of cupiennin 1a, 1a*, 1b, 1d, 1d*, magainin 2, and melittin was determined using fresh human erythrocytes. 1 ml of citrated blood was washed four times with 6 ml of PBS buffer (50 mM sodium phosphate buffer, 150 mM NaCl, pH 7.2) and centrifuged (900 ϫ g) for 6 min at room temperature. The pellet was resuspended in 3 ml and further diluted to a concentration of 1 ϫ 10 9 human erythrocytes/ml in PBS buffer. Lyophilized peptides in various concentrations were resolved in 200 l of PBS buffer, and 50 l of human erythrocytes were added following incubation under gentle shaking at 37°C for 1 h. The samples were then placed on ice and immediately centrifuged at 4°C. The supernatant was carefully removed, and the pellet was resuspended in 240 l of water. Release of hemoglobin was monitored by measuring the absorbance of supernatant and water-treated pellet at 541 nm in a 1-cm cell (Jasco V-550). The negative control (0% hemolysis) was 50 l of human erythrocytes in 200 l of PBS buffer, and the positive control (100% hemolysis) was 50 l of human erythrocytes in 200 l of water. The concentrations of peptide at which 50% hemolysis was observed (EC 50 ) were derived using a sigmoidal curve fitting software (GraphPad Prism 3.0, Graph Pad Software).
Bioassays-Bioassays using Drosophila melanogaster according to Escoubas et al. (31) were performed to estimate the LD 50 (24 h postinjection) of the peptides. For each assay 20 flies were used as control (injecting 0.05 l of insect ringer) and 20 for each of the three peptide concentrations. LD 50 stands for the lethal dose (50% of the test flies die of intoxication) and calculations were done as described elsewhere (32).

RESULTS
Purification of Cupiennins-450 l of venom was separated in a five-step protocol that included gel filtration (Fig. 1A), cationic exchange chromatography (Fig. 1B), and successive reversed-phase-HPLC on a nucleosil 300-5 C 4 column (not shown), a nucleosil 120-5 C 18 column (Fig. 1C) and a nucleosil 100-5 C 8 column (Fig. 1D). The retention times of the purified antimicrobial peptides ranged from 18. Sequence Analysis of Cupiennin 1a-Sequence analysis of cupiennin 1a ceased at position 34. To obtain the rest of the sequence, cupiennin 1a was cleaved with chymotrypsin and the peptides generated were separated by RP-HPLC on a nucleosil 120-5 C 18 column (not shown). The five peaks received were identified by ESI-MS, and the C-terminal peptide (residues 29 -35; ESI-MS 845.40 Da, theoretically 846.43 Da) was sequenced. It is noteworthy that glutamic acid was identified as the C-terminal amino acid ( Fig. 2A). Because of the observed molecular mass difference of 1 Da between the theoretical and measured mass of the chymotryptic C-terminal fragment, an amidation of glutamic acid was supposed. Cupiennin 1a* (*, acidic C terminus) was therefore synthesized to confirm the assumed posttranslational amidation in cupiennin 1a. After purification, cupiennin 1a* (ESI-MS 3799.38 Ϯ 0.39 Da, theoretically 3799.58 Da) eluted as a single sharp peak after RP-HPLC on an analytical nucleosil 120-5 C 8 HD column. The retention time differed slightly from the retention time of cupiennin 1a (Fig. 1D), also indicating the assumed chemical modification. The correct sequence of cupiennin 1a* was fur-ther confirmed by amino acid analysis, ESI-MS, and Edman degradation (not shown). The determined amino acid sequence of cupiennin 1a and 1a* agree well with the results of the amino acid composition analyses (Table I).
Sequence Analysis of Cupiennin 1b, 1c, and 1d-Comparative amino acid analysis of cupiennin 1a, 1a*, 1b, 1c, and 1d (Table I) revealed only slight differences in the content of Ala, Ser, Glx, Ile, Val, His, Thr, and Met. Therefore the amino acid sequence determination of cupiennin 1b, 1c, and 1d was performed by comparative tryptic peptide mapping. Peptide separation was carried out by RP-HPLC (Fig. 3, A and B), and the peptides obtained were then identified by ESI-MS. The tryptic peptides of cupiennin 1a and 1a* are identical (Table II) (Fig. 2B). The sequences of all tryptic peptides that differed from cupiennin 1a fragments were later determined by Edman degradation (Table II). The deduced amino acid sequences of cupiennin 1b, 1c, and 1d (Fig. 2B) are in good agreement with the results of the amino acid analyses (Table I). For further studies a cupiennin 1d analogue with Gln was synthesized as nonamidated C terminus (cupiennin 1d*). The correct sequence of cupiennin 1d* was also confirmed by amino acid analysis (Table I)  General Structural Features-Cupiennin 1a, 1b, 1c, and 1d are linear peptides consisting of 35 amino acid residues. The C terminus is amidated, and the net positive charge is at least ϩ 8 at neutral pH (Fig. 2B). The theoretical isoelectric point is 11.30 for all peptides, not taking the C-terminal amidation into account. The N-terminal part of the sequences (Gly-Phe-Gly- Ala/Ser-Leu-Phe) is somewhat hydrophobic, whereas polar amino acid residues predominate in the C-terminal region. All cupiennins are characterized by six repeats of four amino acids, which form the central part of the peptide chain, following the N-terminal hydrophobic stretch with lysine at every first position in their central part. These six repeats are defined by the following sequence: position 1 is always lysine; position 2 is variable (hydrophobic, charged, or polar amino acid); position 3 is always a hydrophobic amino acid (Leu, Val, Ala, Ile) or Gly; and in position 4 is Ala or Val (Fig. 2B). Based on the consensus scale of hydrophobicity for the individual amino acid residues of Eisenberg (33) the mean hydrophobicity (H) of the cupiennins was found to range between Ϫ0.138 and Ϫ0.168. Assuming an ␣-helical conformation, their hydrophobic moment () was determined to vary between 0.0121 and 0.0282 (Fig. 4A, Table III). These parameters differ distinctly from H and values characterizing other antimicrobial peptides such as lycotoxins (24) isolated from spider venom, melittin (34) from bee venom, and magainin 2 (35) found in frog skin. However, the mean hydrophobicity as well as the hydrophobic moment of the cupiennins are distinctly lower than the H and of melittin (H ϭ Ϫ0.086, ϭ 0.2244), magainin 2 (H ϭ Ϫ0.036, ϭ 0.2861), and lycotoxin I (H ϭ Ϫ0.083, ϭ 0.0681) ( Table III). The angle subtended by polar residues (⌽) describing the hydrophilic helix surface is at 220°unambiguously greater than the polar face of most other helical antimicrobial peptides (Fig.  4A). The amphipathic motif becomes obvious in the helix net projection. The surface of the cupiennin helix is characterized by a right-handed ribbon of positively charged (lysine) side groups and polar amino acids winding around the ␣-helix (Fig.  4B).
Circular Dichroism Spectroscopy-The CD spectra of the peptides were recorded in sodium phosphate buffer and exhibit an unordered peptide structure (Fig. 5a). The addition of TFE induced pronounced spectral changes (Fig. 5, b and c). The negative bands at 207 and 222 nm and the positive ellipticity below 200 nm are characteristic of an ␣-helical conformation. Therefore all peptides were found to be completely helical in the TFE/buffer (1/1 v/v) mixture (Fig. 5). Following Lehrman et al. (36), who suggested that TFE-induced helicity of peptides is a measure of their helix propensity, we take the high ␣ of the investigated cupiennins as an indication of their very high capacity to assume a helical conformation (Table III).
Antibacterial Effects-The cupiennins are highly active against bacteria. Interestingly all four tested bacteria species were susceptible to cupiennin 1a, 1a*, 1d, and 1d* in the nanomolar to micromolar concentration range (Table IV). No clear differences in the minimal inhibitory concentrations were observed between the amidated natural or C-terminal free synthesized cupiennins and E. coli (0.31-0.63 M; cupiennin 1a and 1a*) and E. faecalis (2.5-5.0 M cupiennin 1a and 1a*; 1.25-2.5 M cupiennin 1d and 1d*). The activity of cupiennin 1a (more active) and cupiennin 1a* against P. aeruginosa and S. aureus and of cupiennin 1d and 1d* against P. aeruginosa and S. aureus differed by only one dilution step, whereas the bactericidal effect of cupiennin 1d (more active) and 1d* against E.

FIG. 2. Amino acid sequences of cupiennins from the venom of the spider C. salei.
A, amino acid sequence of cupiennin 1a acquired by sequence analysis until position Met-34 and the chymotryptic peptide 29 -35. B, overview of amino acid sequences of cupiennin 1a*, 1a, 1b, 1c, 1d, and 1d* deduced from a combination of tryptic peptide mapping and sequence analysis of nonidentical tryptic peptides.

TABLE I Amino acid analyses of natural and synthesized (*) cupiennins
The amino acid composition of cupiennin 1a, 1b, 1c, and 1d from C. salei venom purified by RP-HPLC and synthesized cupiennin 1a* and 1d*. The values in parentheses are those calculated from the amino acid sequence. ND, not determined.
7.0 (7) 7.5 (7) 6.5 (7) 6.5 (7) 6.7 coli differed by two dilution steps. Pronounced differences in the susceptibility of Gram-positive and Gram-negative bacteria could not be found. E. faecalis exhibited the weakest suscepti-bility against the cupiennins. Here, melittin was one dilution step more active. However, in the same test system magainin 2 showed no growth inhibition of the tested bacteria up to a FIG. 3. Tryptic peptide mapping of cupiennins from the venom of the spider C. salei. A, after digestion of cupiennin 1a, 1b, 1c, and 1d (5 g) with trypsin (0.5 g), peptides were fractionated isocratically (0.1% trifluoroacetic acid in water) on a nucleosil 120-5 C 18 column. B, in a second stage, further separation was obtained using a linear gradient (0.23% acetonitrile/min) as described under "Experimental Procedures." The separated peptides (F) were identified by ESI-MS (monoisotopic masses, Da) and compared with peptides of cupiennin 1a. Nonidentical peptides (F*) were sequenced by Edman degradation.

and 1d and identification by ESI-MS
Cupiennins were incubated with trypsin (10:1) for 4 h at 25°C. Tryptic peptides were isolated by RP-HPLC (Fig. 3), the molecular mass measured by ESI-MS, and the amino acid sequence determined by Edman degradation. The numbers in parentheses are theoretical values (monoisotopic, Da). The replacement of the amino acids in cupiennin 1b, 1c, and 1d is mentioned in the legend for Fig. 2 (Table V). Compared with melittin (EC 50 , 1.7 M) lytic activity was reduced by a factor of 8.5 for cupiennin 1d* and by a factor of 14.4 and 12.1 for cupiennin 1a and cupiennin 1a*. At a concentration of 8 M, cupiennin 1b induced 15% hemolysis and is thus about twice as active as cupiennin 1a and less effective than cupiennin 1d (30%). Magainin 2 showed no hemolytic effect.
Insecticidal Effects-We also investigated insecticidal effects in a bioassay with D. melanogaster. Surprisingly, cupiennin 1a, 1a*, 1b, 1d, and 1d* showed LD 50 concentrations between 4.7 and 7.9 pmol/mg fly measured after 24 h (Table  V). Differences between the LD 50 doses of the synthesized cupiennins 1a* and 1d* and the natural forms were marginal. Obviously, the neurotoxic effects are independent of C-terminal amidation. In comparison with melittin (14.6 pmol/mg fly) the toxicity of the cupiennins was 2.3-3.1 times higher, and compared with magainin 2 (123.1 pmol/mg fly), it was 19.2-26.2 times higher ( Table V). Injection of peptides at sublethal doses paralyzed the flies in most cases within the first 3 min. Dose-dependent recovery from paralysis was observed within 1-6 h.

Cupiennins as Constitutive Components in Venom-
The venom glands of C. salei store and secrete a highly active hyaluronidase, neurotoxic/cytolytic peptides, and low molecular weight substances such as histamine and taurine, which enable the spider to paralyze its prey within a few seconds (25). In this process the single components act synergistically. For example, hyaluronidase and histamine accelerate the toxin transport into the tissue (spreading factor). Similarly the toxicity of CSTX-1, the main neurotoxic peptide in the venom of C. salei, is increased by histamine and taurine (39). A comparable synergistic effect between various neurotoxic peptides has also been described in Agelena aperta venom (38). But what purpose do the cupiennins serve? We know that the concentration of cupiennin 1a (1.2 mM) in the crude venom of C. salei is in the same range as CSTX-1 (1.4 -3.3 mM) (25,32). This is much higher than the concentration of induced antimicrobial peptides (200 M) found in the hemolymph of insects after injury (40). We also know that C. salei has to use its venom very economically (41, 42) because its complete regeneration takes more than 16 days (43). Both hints suggest that the cupiennins do not act exclusively as antimicrobial agents (bactericidal Table IV) but also have an important function as toxin synergists with additional neurotoxic (LD 50 for D. melanogaster, 84-115 M; Table V) and cytolytic effects (hemolysis, EC 50 14.5-24.4 M; Table V). Comparable antimicrobial lycotoxins from the venom of the spider L. carolinensis act in a neurotoxic mode of operation by dissipating voltage gradients across insect muscle membranes and promoting Ca 2ϩ efflux from rat brain synaptosomes (24). It is therefore tempting to speculate that the ␣-helical melittin in the venom of Apis mellifera (bactericidal concentration, 1. Table IV; LD 50 for D. melanogaster, 263 M; hemolysis, EC 50 Table V), and the cupiennins of C. salei act in a general manner on a wide range of different cell membranes. A common basis is their occurrence in the venom together with hyaluronidase. Such venoms have the function of defending the host against very different organisms or quickly paralyzing its victims. This means that the direct neurotoxic effect of the cupiennins and the synergism with other neurotoxic components such as CSTX-1 are the main function of these peptides in spider venom. 3 Additionally, the bactericidal effect of cupiennins protects the spider venom glands from contamination with bacteria as discussed for lycotoxins (24) and for hadrurin (21) from scorpion venoms.
Primary and Secondary Structure-A comparison with other ␣-helical antimicrobial peptides in the SWISS-PROT sequence data base (release 40.3 of November 16, 2001) and TrEMBL (release 18.3 of November 16, 2001) revealed a fairly unique amino acid sequence for the highly cationic peptides of the cupiennin 1 family. The sequence identity (ALIGN, gene stream network server IGH Montpellier) (44) between cupiennin 1a, 1b, 1c, and 1d is very high (88.6 -91.4%) (Fig. 2). Structural similarities with melittin were noticed in the hydrophobic N terminus and the polar C terminus. The central part of all cupiennins is formed by a remarkable 6-fold repeat of a quadruplet comprising: 1) Lys, 2) variable (charged or polar or hydrophobic), 3) hydrophobic or Gly, and 4) Ala or Val. Cupiennins were found to share some structural similarities (a lysine motif in the central part) with the antimicrobial toxins gaegurin 1, 2, 3, and 4 from the frog Rana rugosa (45) and with lycotoxins from the spider L. carolinensis (24) (Fig. 6). However, gaegurin 3 shows only 34.2% identity with cupiennin 1a, 35.1% with cupiennin 1b/d, and 37.8% with cupiennin 1c. Lycotoxin I shares 31.4% identity with cupiennin 1a, 1b, 1c, and 1d, and the sequence of lycotoxin II is 28.6% identical with that 3 L. Kuhn-Nentwig, unpublished results.

TABLE III Structural properties of cupiennins and other antimicrobial peptides
The net charge of the peptides was calculated under the assumption that under physiological conditions Lys, Arg, and the N-terminal NH 2 are positively charged, and Glu and the C-terminal COOH are negatively charged. The His residues were calculated as not charged. H and were calculated on the basis of the Eisenberg consensus scale of hydrophobicity (33). The percentage of helicity (␣) in sodium phosphate buffer/TFE ϭ 1:1 (v/v) of the peptides was determined from the molar ellipticity at 222 nm according to Chen (29). ND, not determined; *, not amidated. of cupiennin 1a, 1b, and 1c, and 29.7% identical with that of cupiennin 1d. Thus, the structural peculiarities of cupiennins appear to be unique. According to the secondary structure prediction method of Garnier et al. (46), all cupiennins show a high probability of forming an ␣-helix. The enhanced helicity of cupiennin 1b, 1c, and 1d can be associated with the replacement of Val-29 -Val-30 in cupiennin 1a by residues of higher helix propensity (Val-29 -Ala-30 (cupiennin 1d); Ile-29 -Ala-30 (cupiennin 1b and 1c)) (47). CD measurements in the presence of 30 and 50% structure-inducing TFE confirm the high helix forming potential of the peptides.
Many antimicrobial and hemolytic peptides are able to form highly amphipathic helices, and their structural motifs have been described as effective modulators of activity (48). Thus, the magainin helix is characterized by a polar face dominated by the cationic residues that are spread over the whole peptide chain and a hydrophobic surface. H and are moderate (Table III). In contrast to magainin, the melittin helix is determined by the extensive hydrophobic N-terminal peptide domain, whereas the C-terminal part of the helix cylinder is highly cationic (48). The helical wheel projection of cupiennin 1a (Fig. 4A) did not show a distinct separation of polar and hydrophobic surface domains. However, pronounced clustering became obvious in the helix net projection (Fig. 4B). The surface of the cupiennin helix is characterized by a charged spiral shaped ribbon of at least six lysine residues and polar groups that connects the more hydrophobic N-terminal with the polar C-terminal domain. This spiral motif of lysine residues determining the helix surface of the cupiennins is to our knowledge unique for antimicrobial peptides from arthropods. Bactericidal and Hemolytic Activities-A comparison of the MIC values reported here with results obtained by other authors is rather difficult, because different assay conditions (plate versus liquid growth assay) and bacterial strains were used. For reasons of standardization ATCC strains were used, and the MIC experiments were performed as reported by Wu and Hancock (30). The minimal inhibitory concentrations of the most active cationic antimicrobial peptides have been reported to range between 0.25 and 4 g/ml (2). The tested cupiennins are highly active against both Gram-negative and Gram-positive bacteria as shown by MICs ([a] in Table IV) between 0.3 g/ml for cupiennin 1d against E. coli and up to 9.5 g/ml for cupiennin 1a/1a* against E. faecalis. The MIC range of cupiennins contains six dilution steps in contrast to three in the case of melittin. This may indicate differences in the mode of action. Studies of a variety of natural peptides and their chemically FIG. 5. CD spectra of cupiennin 1a, 1a*, and 1d*. a, CD characteristics of cupiennin 1a (C ϭ 5 ϫ 10 Ϫ5 M) in buffer (5 mM sodium phosphate, 150 mM sodium fluoride, pH 7.25) (--) and in TFE/ buffer mixtures: TFE , 30% by volume (⅐⅐⅐⅐); TFE, 50% by volume (---). b and c, CD spectra of cupiennin 1a (--), 1a* (⅐⅐⅐⅐), and 1d* (---) in TFE/buffer mixtures.
[⌰] is the mean residue ellipticity.  modified analogs demonstrate that the cationic charge is essential for recognition and accumulation on the negatively charged bacterial membranes (2,49,50). With their high content of positively charged lysine residues, the cupiennins fulfill this prerequisite. To possess antimicrobial activity, less hydrophobic peptides should have a well developed hydrophobic domain providing the peptide with a highly hydrophobic moment, whereas peptides with low amphipathicity may display antimicrobial activity if they possess an appropriately high intrinsic hydrophobicity. Interestingly, the H and values of the cupiennins are rather low. Thus, either the hydrophobicity of the N-terminal chain region (comparable with melittin) is sufficient to induce membrane disruption, or the high activity is related to specific structural transformations of membranebound peptides based on their unique distribution of charged and hydrophobic residues. A well developed amphipathic helix and high peptide hydrophobicity mediate the hemolytic effect (37,51,52). The activity is determined by hydrophobic interactions between the nonpolar amino acid residues and the hydrophobic core of the lipid matrix of red blood cells (53). The cupiennins are composed of 46 -49% hydrophobic amino acids and show high helicity in the presence of TFE, but structural motifs favoring hydrophobic interactions such as high hydrophobicity and a pronounced hydrophobic moment are poorly developed. H and values are much lower in the cupiennins than in the highly hemolytic melittin and even lower than in the nonhemolytic magainin 2. However, the hydrophobicity of the N-terminal stretches of cupiennins (1.81-2.32) and melittin (2.37) are almost identical. Thus, comparable with melittin, insertion of the hydrophobic N-terminal helix domain into the hydrophobic core of the neutral lipid matrix of red blood cells seems to be responsible for the hemolytic effects of the cupiennins. Additional ionic interactions between the negatively charged O-glycosidic chains of glycophorin A of erythrocytes (54) and the highly cationic lysine ribbon of the cupiennin peptides might support membrane insertion followed by bilayer disturbance, pore formation or lysis. The hypothesis is supported by Helmerhorst et al. (55), who suggested that electrostatic interactions govern the peptide-erythrocyte interaction.
In summary, the cupiennins have been found to be a class of antimicrobial, hemolytic, and bactericidal peptides with unique structural properties. Additional studies are in progress to further elucidate the structural basis of the diversity in their biological activity. Furthermore, the peptides appear to be valuable as model compounds for the investigation of the structural background of membrane lysis and selectivity to different target membranes. FIG. 6. Sequence alignment of cupiennin 1a and 1d compared with amino acid sequences of other antimicrobial peptides. Lysine motifs are boxed, and identical residues are shaded gray. Amino acid differences of cupiennin 1a to cupiennin 1d and identities with other peptides are shaded in dark gray. The asterisks (*) label C-terminal amidation.