The Vertebrate Peptide Antibiotics Dermaseptins Have Overlapping Structural Features but Target Specific Microorganisms*

The physiological significance of the occurrence of se- quence similar antimicrobial peptides in frog skin, as the bombinins in Bombina, the magainins in Xenopus, and the dermaseptins in Phyllomedusa, is a major unan-swered question. Dermaseptins sl, s2, s3, s4, and s5, a family of cationic (lysine-rich), amphipathic antifungal peptides of 28-34 residues were thus synthesized, puri- fied to homogeneity, and evaluated for their growth-inhibition activity in vitro against various pathogenic microorganisms. Although all five of these peptides shared a similar spectrum of lytic activity against the filamentous fungi that are responsible for opportunistic lethal infections that follow the immunodeficiency syn- drome or the use of immunosuppressive agents, they exhibited marked differences in their potencies to ar- rest the growth of Gram-positive and Gram-negative pathogenic bacteria and yeasts. Likewise, whereas der- maseptins sl and s5 were devoid of hemolytic activity, dermaseptin s4 caused lysis of erythrocytes at micromo- lar concentrations. The dermaseptins exhibited dra-matic synergy of action upon combination, resulting in some cases in a 100-fold increase in antibiotic activity of the mixture over the activity of the peptides separately. Shortening the peptide chain of dermaseptin 93 to dermaseptin s3-(1-16)-NH2 did not affect the antimicrobial

The physiological significance of the occurrence of sequence similar antimicrobial peptides in frog skin, as the bombinins in Bombina, the magainins in Xenopus, and the dermaseptins in Phyllomedusa, is a major unanswered question. Dermaseptins sl, s2, s3, s4, and s5, a family of cationic (lysine-rich), amphipathic antifungal peptides of 28-34 residues were thus synthesized, purified to homogeneity, and evaluated for their growthinhibition activity in vitro against various pathogenic microorganisms. Although all five of these peptides shared a similar spectrum of lytic activity against the filamentous fungi that are responsible for opportunistic lethal infections that follow the immunodeficiency syndrome or the use of immunosuppressive agents, they exhibited marked differences in their potencies to arrest the growth of Gram-positive and Gram-negative pathogenic bacteria and yeasts. Likewise, whereas dermaseptins sl and s5 were devoid of hemolytic activity, dermaseptin s4 caused lysis of erythrocytes at micromolar concentrations. The dermaseptins exhibited dramatic synergy of action upon combination, resulting in some cases in a 100-fold increase in antibiotic activity of the mixture over the activity of the peptides separately. Shortening the peptide chain of dermaseptin 93 to dermaseptin s3-(1-16)-NH2 did not affect the antimicrobial potency of the peptide. Further reduction of the chain length yielded peptide derivatives gradually showing reduced activity. Surprisingly, however, analogs of dermaseptin s3 as shorter as 10-12 residues in length remained fully active against Enterococcus faecalis, Cryptococcus neoformans, and against Aeromonas caviae, the causal agent of red-leg disease in amphibians. Overall, these results suggest that, despite 40% sequence similarities, the dermaseptins have distinct spectra of antimicrobial activity and may act in concert to circumvent host invasion by providing frogs with a better shielding against a broad array of microorganisms. They also demonstrate the potential usefulness of short analogs of these peptides as potential candidates for biorational design of germicides. I t is increasingly acknowledged that besides the highly specific cell-mediated immune system, vertebrates are endowed with an additional chemical defense system made up of distinct families of broad-spectrum antimicrobial peptides (1)(2)(3)(4) that are analogous to those found in insects (5,6). Much of our current knowledge concerning this peptide-based immune sys-* This work was supported in part by funds from the CNRS, the INSERM (CRE93), and la Ligue Nationale contre le Cancer. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tem is drawn from the studies on antimicrobial properties of peptides isolated from the skin and the gastrointestinal system of amphibians that take part in the defensive response of frogs against invading microorganisms. These peptides, as the bombinins from the European toad Bombina variegata (7, 81, the magainins from the African clawed frog Xenopus Zaevis (g-ll), and the dermaseptins from the South American arboreal frog Phyllomedusa sauvagii (12, 131, vary considerably in chain length, hydrophobicity, and overall distribution of charges, but have the common feature of being linear cationic (lysine-rich) structures that can adopt a n amphipathic a-helical conformation upon association with lipid bilayers. The available experimental evidence concerning the mode of action of these peptides (14)(15)(16)(17)(18)(19)(20)(21)(22)(23) suggest that their membrane lytic action is mediated by direct interactions with the lipid groups of the microbial membranes leading subsequently to alteration of the membrane functions responsible for osmotic balance. Unlike a-helical polycations such as polylysines that have long been known to possess antimicrobial and hemolytic properties (24), most of the naturally occurring antimicrobial peptides from frog skin have little or no hemolytic activity, indicating that the basicity and overall hydrophobicity of the molecule are important factors governing selective membrane recognition.
All the amphibian species examined to date have been shown to synthesize their own set(s) of antimicrobial peptides, constituting families of 2-11 closely related members that often differ with but a few amino acid substitutions. Other examples of natural defensive peptide families include the defensins from circulating leukocytes (2,25) and insect cecropins (5). Although redundancy within peptide sequences has been proposed to be a prerequisite for a role in defense (4,261, the physiological significance of the occurrence of numerous sequence similar peptides within families of antimicrobial peptides has not been demonstrated. In that regard, antimicrobial peptides belonging to the demaseptin family provide a useful tool to assess whether the presence in frog skin of multiple forms of a prototypical peptide sequence, acting separately or in concert, has a selective survival value in habitats laden with microorganisms. Dermaseptin s l , a 34-residue peptide isolated from the skin of the arboreal frogl? sauvagii (121, is the prototype of this family of peptides. It is unique among the antimicrobial peptides in that it has lytic activity against bacteria, yeasts, protozoa (221, and the filamentous fungi that are responsible for opportunistic lethal infections that follow the immunodeficiency syndrome or the use of immunosuppressive agents (27). Until now, four additional congeners are known (Table I), dermaseptins s2, s3, s4, and s5 (13), which share high levels of sequence similarities with dermaseptin s l (53-94% amino acid positional identity), suggesting that their respective genes are all members of the same family. These peptides vary in length from [28][29][30][31][32][33][34] residues, contain between 3 and 6 lysine residues, and have amino acid sequences exhibiting the periodic pattern of a-helixes containing sharply demarcated polar and nonpolar faces. All five of of Dermaseptin Peptides -** -* -*** -* --1-28

1-28
Sequences alignment was performed by using Clustal V software (30) with fured gap penalty. Identical (*I and similar (-) residues among the sequences are indicated. these peptides adopted random conformation in dilute aqueous solutions, but could be induced to form helical amphipathic structure in the presence of apolar solvents (13). These peptides differ in their chain lengths, charge ratio, and hydrophobicity, thus allowing the effect of these parameters on peptide activity and selectivity to be investigated. Toward this goal, we have synthesized the five dermaseptins as well as a series of analogs and evaluated their ability, either alone or in combination, to inhibit the growth of various pathogenic agents. The results reported herein suggest that the simultaneous presence of the five dermaseptins provides frogs with a better shielding against a wider range of infecting microbes. They also demonstrate the potential usefulness of these peptides as candidates for biorational fungicides or models for the synthesis of such fungicides.

MATERIALS AND METHODS
Solid Phase Peptide Synthesis-Peptides were prepared by stepwise solid phase synthesis using Fmoc' polyamide-active ester chemistry on a Milligen 9050 pepsynthesizer. All Fmoc-amino acids were from Milligen. 4-(hydroxymethy1)Phenoxyacetic acid-linked polyamidelkieselguhr resin (pepsin KA), Fmoc-amino acid pentafluoro phenyl (F'fp), and 3hydroxy-2,3-dehydro-4-oxo-benzotriazine (Dhbt) esters were from MilligenEiioresearch. Side chain protection were tert-butyl for Ser and Thr, trityl for Asn and His, t-butoxycarbonyl for Lys, and oxygen tertbutyl for Glu and Asp. Cleavage of peptidyl-resin and side chain deprotection were carried out at a concentration of 5 mg of peptidyl-resin in anisol, water, and ethyl methyl sulfide (82.5, 5 , 5 , 5, and 2.5 v/v) for 2 h 1 ml of a mixture composed of trifluoroacetic acid, para-cresol, thioat room temperature. After filtering to remove the resin and ether extraction, the crude peptides were purified by a combination of Sephadex gel filtration, ion exchange chromatography, and preparative HPLC. Homogeneity of the synthetic peptides was assessed by analytical HPLC, amino acid analysis, solid phase sequence analysis, and mass spectrometry as described elsewhere (12).
Biological Assays-Antimicrobial assays were performed in sterilized 96-well plates (Nunc F96 microtiter plates, Denmark) in a final volume of 100 pl as follows. 50 pl of suspension containing lo6 spores or microconididml in Sabouraud glucose broth or lo6 bacteridml in Luria-Bertani (LB) culture medium were incubated in the presence of serial 2-fold dilutions of the synthetic dermaseptins in 50 pl of water, or in the presence of 50 pl of 0.4% formaldehyde in water as a negative control, or in the presence of 50 pl of water as a positive control. The synthetic peptides were weighed in a microbalance and solubilized in water at the desired primary dilution. Inhibition of growth was determined by measuring optical density at 492 nm with a Titertek Multiskan MCC after an incubation time of 24 h a t 30 "C (37 "C for bacteria). Each minimal inhibitory concentration (MIC) was determined from two independent experiments performed in duplicate.
Hemolytic activity of the synthetic peptides was assayed with heparinized fresh human blood rinsed three times with phosphate-buffered saline by centrifugation for 15 min a t 900 x g. Red blood cells (108/ml) were then incubated under agitation at 37 "C in distilled water for 100% The abbreviations used are: Fmoc, N-(9-fluorenyl)methyoxycarbonyl; HPLC, high performance liquid chromatography; LB, Luria-Bertani; MIC, minimal inhibitory concentration.
hemolysis, in phosphate-buffered saline (50 m~ sodium phosphate, 150 mM NaC1, pH 7) for control, or in phosphate-buffered saline containing various concentrations of the peptide (up to 0.250 mg/ml) in a final volume of 0.2 ml. Release of hemoglobin was monitored after centrifugation at 900 x g by measuring the absorbance of 100 pl of supernatant at 541 nm after 1 and 24 h of incubation. Part of the supernatant was dispatched for HPLC analysis as described below.
Reversibility of growth inhibition was assessed as follows. Peptides with a linear gradient 0 4 0 % of acetonitrile containing 0.07% trifluoroacetic acid. This procedure was also applied to verify peptide stability after 1 h of incubation with red blood cells.

RESULTS
Spectrum ofAntimicrobia1 Activity of the Dermaseptins-The synthetic replicates of dermaseptins s l , s2, s3, s4, and s5 were investigated for their ability to affect the viability of various prokaryotic and eukaryotic cells in culture media. Their ability to induce cytolysis or to inhibit cell proliferation is reported in terms of MIC, defined as the dose a t which 100% inhibition of growth was observed after 24 h of incubation at 30 "C (37 "C for bacteria). As shown in Table 11, each dermaseptin revealed to be endowed with a large spectrum of antimicrobial activity, including Gram-positive and Gram-negative actinomycetales bacteria, cocci, and rods, as well as yeasts and filamentous fungi. Interestingly, the dose-response profiles obtained showed sharp curves in which 0-100% inhibition was generated within a 2-3-fold peptide dilution (Fig. 1). However, a close inspection of the tabulated values reveals complex patterns of antimicrobial potencies. Indeed, despite extensive structural similarities between the dermaseptin family members, they exhibited marked differences in their efficacy to inhibit microbial proliferation. These differences are most pronounced with bacteria. For instance, whereas dermaseptins s l , s2, s3, and s4 showed high potency against the causal agent of "red-leg" disease in amphibians, A. cauiae (MIC = 0.5-1 m), or against E.  Conversely, dermaseptin s5. Although to a lesser extent, these differences in potency s5 turned out to be the most efficient peptide in inhibiting the were also found against the yeasts Cryptococcus neoformans, C. proliferation of the Gram-positive cocci S. aureus (MIC = 2 p~) albicans, and Saccharomyces cereuisiae (Table 11). In sharp concompared with dermaseptins s l , s2, s3, and s4 (MIC = 5,20,10, trast, all the five peptides were almost as efficient in inhibiting and 10 p~, respectively). On the other hand, the order of PO-the growth of pathogenic fungi at concentrations ranging from tency against Nocardia brasiliensis, a Gram-positive filamen-10 to 40 p~ (Table 11). The only noticeable exception being tous branching bacteria, was dermaseptins s3 > s4 >s2 > sl > dermaseptin s5 which showed reduced potency against the fil- Nocardia brasiliensis (IP16-80) 5 1 Saccaromyces cerevisiae (IP118079) 5 Candida albicans (IP884-65) The MIC is defined as the dose at which 100% inhibition of growth was observed after 24 h of incubation in culture media.
amentous mold Aspergillus fumigatus. Another noticeable difference is evidenced by the ability of the dermaseptins to interact with human red blood cells. As reported in the lower part of Table 11, 1 h treatment with dermaseptins s l or s5 did not permeate the erythrocytes up to the highest concentration assayed (70 and 90 p, respectively). Hemolysis of erythrocytes however occurred after treatment with dermaseptins s2 or s3 at 70 and 80 PM, respectively. Under the same conditions still, treatment with dermaseptin s4 resulted in 100% hemolysis at 1 PM (50% hemolysis at 0.5 PM).
The Dermaseptins Exhibited Synergy of Action upon Combination-To investigate a possible synergetic relation between the dermaseptins, each peptide was tested in serial 2-fold dilutions against various microorganisms in the presence of a constant equimolar mixture composed with the remaining four peptides. As an example, Fig. 1 shows typical dose-dependent curves of growth inhibition obtained for A. fumigatus or N. brasiliensis after 24 h of treatment by the individual dermaseptins. In the absence of the other family members, the peptides exhibited MIC values ranging between 20 and >70 PM for A, fumigatus, and between 5 and 40 1 " for N. brasiliensis.
However, in the presence of a mixture composed of 0.25 p of each of the dermaseptins sl, s2, s4, and s5, the MIC of dermaseptin s3 dropped from 20 to 3 PM and from 5 to l p~, respectively (Fig. 1). Note that, individually, none of the five dermaseptins is active at l p against either one of the two microorganisms. This potency enhancement was observed for dermaseptin s3 (Table 111) as well as for the other dermaseptins (data not shown) with various microorganisms. As shown in Table 111, synergy of action upon combination resulted in some cases in a 100-fold increase in antibiotic activity of the mixture over the activity of the peptides separately.
Short Dermaseptin Analogs Are Highly Potent against Several Microorganisms-To evaluate the structural features responsible for the antimicrobial activity of the dermaseptins, COOH-terminally truncated fragments of dermaseptin s3 were assayed against various microorganisms. The data collected are summarized in Table IV Neutralization of the negative charge at the COOH terminus of dermaseptin s3 by conversion of the carboxylate to a carboxamide yields a peptide derivative exhibiting enhanced antimicrobial potency (2-10-fold) over a broad array of pathogenic microorganisms. Stepwise elimination the COOH-terminal 14 residues of dermaseptin s3 to give dermaseptins s3-(1-20)-NH2 and s3-(1-16)-NH2 did not significantly alter the peptide potency. For instance, the 16-residue peptide retained many of the features of the parent compound and conserved a large spectrum of activity. Only in few cases, such as with S. aureus and I ? . brasiliensis, these shorter analogs showed reduced potency compared with the parent molecule. Interestingly, although highly potent against pathogenic microorganisms, s3-(1-16)-NH2 was devoid of hemolytic activity even at higher concentrations (150 p) compared with dermaseptin s3 ( Table Iv). Further stepwise shortening of the chain length down to dermaseptin s3-(1-10)-NH2 yielded peptide derivatives exhibiting a slow gradual loss in potency compared to dermaseptin s3, namely, against S. aureus or N. brasiliensis as well as against the filamentous fungi. Surprisingly, however, these shorter analogs still displayed high potency against bacteria, such as Aeromonas cauiae, Escherichia coli, Enterococcus faecalis, and the yeasts. In few cases, s3-(1-10)-NH, was practically as efficient as the native peptide, yet this shorter s3 version was devoid of hemolytic activity.
Stability of the Dermaseptins-To evaluate the susceptibility of peptides to eventual enzymatic degradation, aliquots from the suspensions assayed with dermaseptins s3, s3-(1-16)-NH2 and s3-(1-10)-NH2 were analyzed by HPLC at various incubation times against C. albicans, A. caviae, S. aureus, and human erythrocytes. Whereas dermaseptin s3 and s3-(1-16)-NH2 displayed potent inhibitory activity against S. aureus, their concentrations in the suspensions gradually decreased in time and vanished after 24 h of incubation (Fig. 2). While devoid of antimicrobial activity against S. aureus, dermaseptin s3-(1-10)-NH2 was conserved almost intact in the suspensions after 24 h of incubation time. On the other hand, although these three peptides were active against C. albicans and A. cauiae, their concentrations were stable throughout the experiment as at least 90% of the initial peptide could be recovered after 24 h of incubation (data not shown). While the disappearance of the active peptides may result from differential susceptibility to intracellular proteolytic enzymes released consequently of their cytolytic activity, these results suggest that there may not be necessarily a direct correlation between peptide stability and activity. In addition, they clearly rule out the possibility that the lack of activity is due to enzymatic degradation. Therefore, the observation that dermaseptin s3-(1-10)-NHZ is devoid of antimicrobial activity against S. aureus may mostly reflect the lack of the molecular elements that are responsible for membrane recognition and activity. The fact that dermaseptin s3, a hemolytic peptide, is partially degraded after 1 h of incubation with human erythrocytes while the nonhemolytic peptides dermaseptins s3-(1-16)-NHZ and s3-(1-10)-NHZ are stable is in accordance with this scheme (data not shown).
Dermaseptins Cause Irreversible Growth Inhibition-To gain insight into the mechanism of action of the dermaseptins, suspensions of C. albicans, A. caviae, and s. aureus were treated each with 0.2 mg/ml of dermaseptins s3, s3-(1-16)-NH2 or s3-(l-10)-NH2. After various incubation times, the microorganisms were harvested by centrifugation, thoroughly washed and reincubated in fresh medium for 24 h. M e r 10 min of treatment with any one of the three peptides, washed C. albicans and A. caviae did not proliferate. Whereas washed S. aureus did not proliferate when treated previously with dermaseptins s3 or with s3-(1-16)-NH,, it did proliferate when treated with dermaseptin s34 l-lO)-NHZ. These results remained unchanged after 1 or 24 h of treatment. This correlates nicely with the results reported in Table IV and demonstrates that the effects of the dermaseptins are instantaneous and irreversible.

DISCUSSION
In the attempt to understand possible reasons for the occurrence of many structurally related dermaseptins in frog skin, dermaseptins s l , s2, s3, s4, and s5 were prepared by stepwise solid phase synthesis, purified to homogeneity, and investi-  The MIC is defined in the legend of Table 11, a = amide. 100% hemolysis after 1 h of incubation. Not determined.

Growth inhibition activity in vitro of dermaseptin s3 and truncated analogs against bacteria, yeasts, and fungi
gated for their ability to affect the viability of various prokaryotic and eukaryotic cells in vitro. Overall, the data presented show for each peptide broad, but not identical, spectra of antimicrobial activity. At micromolar concentrations, each dermaseptin was highly efflcient in killing selectively certain microbes (Table 11). Moreover, their activity was considerably amplified upon combination. Hence the biological significance of the presence of five closely related dermaseptins in frog skin may thus be understood in terms of advantage for survival under hostile environmental conditions. To defend itself against pathogenic microbes, the naked skin of frog would produce a fued repertoire of small-sized and structurally related antimicrobial peptides that are promptly synthesized at a low metabolic cost, easily stored in large amounts, and readily available to provide the frog with a maximum coverage over a wide range of possible invading microorganisms. Concerning the mode of action of the dermaseptins, previous studies performed with dermaseptin s l have suggested that basicity and amphiphilicity of the NH,-terminal domain are responsible for binding and aggregation of the peptide at the plasma membrane, inducing consequently cell lysis (21)(22)(23). Although the precise mechanism of action of the four new dermaseptins is yet to be demonstrated, it is likely that they act in a manner similar to that established for dermaseptin s l (21), i.e. (i) complementarity between the charge distributions of the peptide and the polar head groups of phospholipids initiates association of the peptide with the membrane surface via electrostatic interactions; (ii) upon interacting with the apolar interface, the peptide adopts a-helical structure corresponding to its hydrophobic period that provides a low energy conformation in which the hydrophobic residues are maximally dehydrated and the polar residues are maximally hydrated; (iii) the helical peptide penetrates deeply into the membrane contacting fatty acyl chains and disrupting the fluidity of the membrane. Concordant with this view are the following observations: (a) multiple sequences alignment of the five dermaseptins yielded 40% amino acid positional identity ( Table 1); ( b ) the dose response curves (Fig. 11, obtained with all dermaseptins, evolved from 0 to 100% inhibition generally within the range of 2-3-fold peptide dilutions, suggesting that peptide aggregation occurs a t a threshold concentration (21); and (c) the effect induced by treatment with each dermaseptin was rapid (<lo min) and irreversible.
Since the dermaseptins differ only in their net charge, charge distribution, hydrophobicity of the nonpolar face, and average angle subtended by the polar face (Table I), the effect of these parameters on lipid affinity, and thus on peptide potency, can be investigated (29). As shown in Table I1 however, some dermaseptins exhibited a similar potency against several microorganisms despite the fact that they greatly differ in their above cited structural parameters. On the other hand, despite extensive homology in their structural profiles, some dermaseptins exhibited up to 70-fold differences in potency to inhibit the growth of various microorganisms (Table 11). In a similar vein, one property of the dermaseptins that should be explained is their differential ability to target specific microorganisms. Indeed, the dermaseptins displayed a heterogeneous profile of potencies over the different microorganisms tested (Table 11).
Perhaps, the most remarkable divergence between them is their preferential affinity for pathogenic microbes versus mammalian cells, as most dermaseptins are barely or not toxic for human erythrocytes while dermaseptin s4 is highly hemolytic. A possible explanation for this may lie in the fact that s4 has, by far, the highest hydrophobic moment. When compared with the remaining dermaseptins, the summed hydropathic indexes of its side chains is +28.9 compared with +6.3, +13.0, +10.5, and +9.7, respectively, for dermaseptins s l , s2, s3, and s5 (Table I). Also, the highest difference in hydrophobicity is localized at the NH, terminus, a domain being of crucial importance for effective interactions with membranes (23). Therefore one is tempted to hypothesize that highly hydrophobic NH, terminus within amphiphilic cationic peptides would confer hemolytic activity. This hypothesis is experimentally supported by the use of the 27-residue antimicrobial nonhemolytic peptide, dermaseptin bicolor (DS bl): DVLKKIGTVALHAGKAALGA-VADTISQ-NH, (28). Addition of hydrophilic residues to its NH, terminus resulted in a considerable enhancement of its antimicrobial potency but did not affect its hemolytic properties (data not shown). Further addition of hydrophobic residues yielded peptide derivatives increasingly hemolytic with increasing hydrophobicity i.e. 100% hemolysis was observed at 15 pg/ml for LMWKK-DS b l , at 60 pg/ml for WKK-DS b l ; no hemolysis was observed up to 1000 pg/ml for KK-DS b l or for native DS bl. In that regard, it is interesting to note that a weak hemolytic activity was observed for dermaseptin s2 but not for s l , dermaseptin s2 being slightly more hydrophobic due to two substitutions i.e. WF and D/N at positions 4 and 27, respectively. Accordingly, the hemolytic capacity of dermaseptin s3 (Table IV) may have been abolished by decreasing its hydrophobicity, the summed hydropathic indexes of s3-(1-16)-NH, and s3-(1-10)-NH2 being +0.2 and +3.2, respectively, compared with +10.5 for the native peptide.
Thus, whereas slight structural variations, e.g. the net charge, the charge distribution around the polar face, and bulkiness of the nonpolar face seem important, although not fully understood, factors for fine tuning the lipid targeting of these peptides, it is most probable that lipid affinity may vary considerably depending on the composition of the fatty acyl chains, the content of cholesterol, and the hydrophobic thickness of the lipid bilayers. Therefore, in contrast to the currently growing opinion that often generalize the bioactive properties of cationic-amphiphilic peptides, this study showed that this view may be oversimplified since subtle structural differences in the dermaseptin family members lead to marked differences in both the spectrum of activity and in potency. In a preliminary study that concerned dermaseptin sl (23), we showed that the mere deletion of 2 residues from its NH, terminus lowered its lytic potency against C. albicans and C.
neoformans. On the other hand, dermaseptin s1-(1-18) displayed a comparable potency with that of the native peptide, and dermaseptin ~1416-34) was devoid of antimicrobial activity (23). In addition, amidation of the carboxyl at the COOHterminal end of dermaseptin s1-(1-18) improved its potency. This suggested that the a-helical NH,-terminal domain of dermaseptin s l contains the essential features responsible for its activity. To probe further these properties we evaluated the minimal structural requirements for potent antimicrobial activity using dermaseptin s3. This peptide represents a convenient tool to investigate the structure-function relationships in terms of how the chain length and charge content correlate with lytic potency since, among the five dermaseptins, dermaseptin s3 presents the most regular structure, with lysine residues punctuating every fourth amino acid position along the peptide chain. In agreement with previous observations, we find in this study that amidation of the carboxyl at the COOHterminal end of the native peptide improves its potency (Table  IV). However, analysis of the results obtained with COOHterminally truncated analogs revealed complex patterns of behavior, making it impossible to draw general predictive rules. For the sake of clarity the results obtained from these experiments were divided into two groups (Fig. 3): a group of microorganisms for which stepwise deletions from 30 t o 10 residues gradually led to loss of potency (Fig. 3, upper diagram), and a group for which potency was hardly affected by these modifications. In the first group, deletion of the 10 COOH-terminal residues, which did not vary the net positive charge of the peptide, resulted in a general increase in MIC except for E. faecalis and C. albicans for whom the MIC remained un-changed. Further reductions in chain length down to dermaseptin s3-(1-16)-NH2 which, conserve the net positive charge of the parent peptide, resulted in a further increase in MIC for S. aureus, N. brasiliensis, and C. albicans, while no change in MIC was observed for Aspergillus niger. Interestingly, the chain reduction resulted in a more efficient inhibition of Arthroderma simii and A. fumigatus. Reduction of the chain length down to dermaseptin s3-(1-12)-NW2 reduced the net positive charge of the peptide and led to a further loss of potency for S. aureus and N. brasiliensis and, although to a lesser extent, for C. albicans, A. simii, A. fumigatus, and A. niger which still displayed a good sensitivity. Finally, all members of this group were insensitive to the 10-mer version probably due to the loss of the molecular elements responsible for its activity against these microorganisms. As illustrated in the lower diagram of Fig. 3, all of the mentioned above modifications hardly affected the members of the second group (A. caviae, E. coli, E. faecalis,  C. neoformans, and S. cereuisiae). The MIC observed for all peptides remained practically unchanged, analogs of dermaseptin s3 as shorter as 10-12 residues in length being highly active against these microorganisms. These results demonstrate the potential usefulness of these short analogs as candidates for biorational-targeted antibiotics or as a model for the design of such antibiotics. Finally, in the search of a universal minimal pharmacophore, this study revealed the feasibility of such a task only when it is targeted against individual microorganisms.