The NMR-derived Solution Structure of a New Cationic Antimicrobial Peptide from the Skin Secretion of the Anuran Hyla punctata*

Amphibian skin secretions constitute an important source of molecules for antimicrobial drug research in order to combat the increasing resistance of pathogens to conventional antibiotics. Among the various types of substances secreted by the dermal granular amphibian glands, there is a wide range of peptides and proteins, often displaying potent antimicrobial activities and providing an effective defense system against parasite infection. In the present work, we report the NMR solution structure and the biological activity of a cationic 14-residue amphiphilic α-helical polypeptide named Hylaseptin P1 (HSP1), isolated from the skin secretion of the hylid frog Hyla punctata. The peptide antimicrobial activity was verified against Candida albicans, Staphylococcus aureus, Escherichia coli, and Pseudomonas aeruginosa, whereas no significant lytic effect was detected toward red or white blood cells.

Because of its role in respiration and osmoregulation, the skin of amphibians is made up of a thin layer of cells, densely populated by moisturizing glands, that facilitate the gas exchange necessary for survival of the animal. Despite being a poor mechanical barrier against biotic and physical stress, the epidermis, in particular the "stratum corneum," has evolved a complex mechanism of defense. A wide range of chemically diverse components is rapidly secreted by the granular glands covering the skin. In addition to the psychotropic, toxic, and repellent effects that provide an effective strategy to keep predators away (1), a large number of these secretions also contain potent antimicrobial agents that protect the amphibians against infections by molds, bacteria, and protozoa (2)(3)(4).
As for the antimicrobial agents, it is becoming clear that the cationic peptides represent an important component of this host-defense system, because they are rapidly expressed and secreted soon after the first contact with the pathogens, a process induced by bacterial products (5). The binding of cati-onic peptides to the anionic lipids of biomembranes is the initial step in the permeabilization of the microbes cell wall, either via the formation of pores, involving the "barrel-stave" mechanism, or by the induction of phospholipids "flip-flop," involving the "carpet-like" mechanism (6,7). Indeed, these peptides share the propensity to acquire an amphiphilic ␣-helical structure when in hydrophobic environments, a structural feature essential for the initiation of the interaction with the phospholipid bilayer of the target membrane (8,9), that leads to membrane lysis and, as a final result, to inhibition of growth and cell death.
The indiscriminate use of antibiotics has led to the emergence of highly resistant strains of microorganisms. Because of this, the screening for new naturally occurring antimicrobial agents is constantly growing and being expanded to include the chemicals involved in the defense of animals such as the arthropods and amphibians. The antimicrobial peptides found in amphibian skin secretions have been successfully tested in "in vitro" assays. So far, the Magainins (2), brevinins (10), and dermaseptins (11) are the best known broad-spectrum antimicrobial peptides obtained from frogs. In this work, we report the structural and functional characterization of a 14-residue cationic peptide isolated from the skin extract of Hyla punctata, a hylid frog commonly found mainly in the wet land areas of the Brazilian north and northeast regions.

MATERIALS AND METHODS
HPLC 1 -H. punctata specimens captured in Palmas, State of Tocantins, Brazil (IBAMA license 0637/91 A.C), were mildly stressed by electrical stimulation (around 6 V), and the skin secretion was collected in Milli-Q H 2 O. The crude extract was filtered, lyophilized, and fractionated by RP-HPLC (Shimadzu Co.) using a Vydac 218TP510 column. The fractions were eluted in a 60-min linear gradient of H 2 O (solvent A) and acetonitrile (solvent B), both containing 0.1% trifluoroacetic acid. The experiment was monitored at 216 and 280 nm, and samples were lyophilized and stored at Ϫ80°C.
Mass Spectrometry Analysis and de Novo Sequencing-Aliquots of * This work was supported in part by Fundaçã o de Amparo a Pesquisa do Estado de Sã o Paulo, Brazil, Projects 99/11030-9, 99/07574-3, and SmolBNet and Laboratório Nacional de Luz Síncrotron. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ ¶ To whom correspondence should be addressed. E-mail: cbloch@ lnls.br. 1 The abbreviations used are: HPLC, high performance liquid chromatography; HSP1, hylaseptin P1; RP-HPLC, reversed phase HPLC; Fmoc, N-(9-fluorenyl) methoxycarbonyl; NCCLS, National Committee for Clinical Laboratory Standards; MIC, minimal inhibitory concentration; MFC, minimal fungicidal concentration; TFE, trifluorethanol; NOE, nuclear effect Overhauser; DQF-COSY, double quantum filtered correlation spectroscopy; TOCSY, total correlation spectroscopy; NOESY, nuclear Overhauser enhancement spectroscopy; ROESY, rotating frame Overhauser spectroscopy; r.m.s.d., root mean square deviation; AFM, atomic force microscopy; LPS, lipopolysaccharides. the native and the synthetic peptides in 50 pM aqueous solutions were mass analyzed by matrix-assisted laser desorption ionization-time of flight/mass spectrometry, on an ABI 4700 Proteomics Analyzer with TOF-TOF optics (Applied Biosystems). The samples were mixed with a saturated matrix solution of ␣-cyano-4-hydroxycinnamic acid (1:3) and spotted on a MALDI-TOF/TOF sample plate. The MS and the MS/MS spectra were carried out in the reflector mode with external calibration, using the calibration mixture Sequazyme Standard kit (Applied Biosystems). Peptide de novo sequencing was performed by precursor ion fragmentation using N 2 as collision-induced dissociation gas, and collision cell pressure was kept at 2.8 ϫ 10 Ϫ6 torr.
Edman Degradation-The automatic N-terminal sequencing of HSP1 was performed on a Protein and Peptide Sequencer PPSQ-23 (Shimadzu Co.) following the manufacturer's instructions.
Chemical Synthesis-The synthetic HSP1 was obtained by automatic solid phase Fmoc synthesis on a Pioneer Synthesis System (PerSeptive Biosystems), following the manufacturer's instructions. After chemical deprotection and lyophilization, the peptide was purified by preparative RP-HPLC, followed by analytical RP-HPLC, using Vydac columns (218TP1022 and 218TP54, respectively). The final HPLC fraction had its purity and amino acid sequence mass analyzed by MALDI-TOF/TOF.
Antibacterial Assays-The three bacterial strains used to investigate the antimicrobial activity of the peptide HSP1 were Pseudomonas aeruginosa ATCC 27853, Staphylococcus aureus ATCC 25923, and Escherichia coli ATCC 25992, which were purchased from American Type Culture Collection. Microorganisms cultured at Laboratório SABIN de Aná lises Clínicas/Brasília-DF, Brazil, were grown in stationary culture at 37°C and after that were transferred to Mueller-Hinton liquid medium, according to the National Committee for Clinical Laboratory Standards (NCCLS) to perform the assays (12).
The bacterial liquid growth inhibition assay was performed as described by Bulet et al. (13). The peptide was dissolved up to 8-fold in Mueller-Hinton broth by serial dilution. The highest peptide concentration used for the assay was 256 g/ml (195.2 M) in an initial inoculum of 1.8 ϫ 10 7 colony-forming units/ml. The final volume was 200 l per well, 100 l of the peptide and 100 l of the inoculum. The experiment was carried out in stationary culture at 37°C, and the spectrophotometric readings were performed 12 h after incubation. The minimal inhibitory concentration, as introduced by Park et al. (14), was determined based on three independent measurements, using the optical density parameter (A 595 nm ).
Conventional antibiotics had their minimum inhibitory concentrations determined against the three experimental bacterial strains, by automated biochemical analysis (Vitek, bioMériuex Inc.). To distinguish between bacteriostatic effects and bactericidal ones, the same bacteria species used in the previous experiments were incubated for 12 h in the MICs of HSP1 and plated on solid culture medium containing 1% noble agar. The plates were subsequently incubated and examined after 24 and 48 h (data not shown). All assays were performed in triplicate (15).
Antifungal Assays-The peptide HSP1 was tested against two yeasts isolates, including a clinical Candida albicans strain and one of C. albicans yeast strain from the ATCC. The in vitro antifungal assays were performed using the broth microdilution method, according to the recommendations from the NCCLS (16).
Candida sp. cells freshly grown on slopes of Sabouraud dextrose agar were suspended in 3.7% brain-heart infusion (Difco); the cell concentration was adjusted to 1.5 ϫ 10 6 cells per 1 ml and diluted to 10 4 cells per 1 ml. The solution containing the peptide was added to 96-well plates (100 l per well) and serially diluted 2-fold. The final concentrations of peptide in the wells ranged from 8.0 to 256.0 g/ml (from 6.1 to 195.2 M). After inoculation (200 l per well, 5 ϫ 10 3 cells per ml), the 96-well plate was incubated at 33°C for 48 h, and the absorbance was measured at 595 nm, using an automatic plate reader (Bio-Rad 450 Microplate Reader) (17). MIC of all peptides and drugs was defined as the first concentration of peptide in which turbidity in the well was 50% less than that in the control well (18).
To determine the minimal fungicidal concentrations (MFCs), 50-l samples were withdrawn from the well containing the MIC and from those containing all concentrations above the MIC onto Sabouraud dextrose agar plates. The samples were incubated in quadruplicate. The plates were then streaked with a sterile loop and incubated at 37°C for 48 h. The MFC was defined as the lowest concentration of the peptide in which no growth was detected (19,20).
Hemolytic Tests-The hemolytic activity of HSP1 was determined using fresh human blood from a healthy O ϩ donor, collected on heparin (final concentration 24.2 units/ml), and centrifuged at 980 ϫ g for 3 min at room temperature. The test was approved by the Medical Ethics Committee of the Laboratório SABIN de Aná lises Clínicas and with consensus of all individual donors. The pellet of erythrocytes was washed three times in saline buffer (20 mM Tris, 130 mM NaCl, pH 7.4) and suspended in the same buffer to obtain 10 7 human erythrocytes/ml (OD ϭ 2.0; ⑀ 414 ϭ 4.2 ϫ 10 5 M Ϫ1 cm Ϫ1 ) (21). Aliquots of HSP1 were serially diluted from 195.2 to 6.1 M (256 to 8 g/ml) in 50 l of saline buffer, and 150 l of the erythrocyte suspension was added, followed by incubation under gentle shaking at 37°C for 30 min. The samples were then placed on ice and immediately centrifuged at 980 ϫ g and at 4°C (23). The absorbance of the supernatants was measured at 414 nm, and the total hemolysis reference was obtained by the addition of distilled water on the erythrocyte pellet (21).
Cytometry-The peptide HSP1 was dissolved in 100 l of saline buffer and incubated with human blood plasma (900 l), giving a 195.2 M peptide solution. Triplicates of 50 l were collected after 20 and 240 min of incubation and analyzed by a Cell-Dyn 3500 SC/SL System (Flow Cytometry Automated Hematology Analyzer, Abbott), in order to obtain the blood cell total count and the mean platelet volume evaluation, after exposure to HSP1. The negative control was obtained using the saline buffer (22).
CD Spectroscopy-Circular dichroism experiments were performed using a Jasco J-810 spectropolarimeter (Jasco International Co. Ltd., Tokyo, Japan), coupled to a Peltier Jasco PFD-425S system for temperature control. Samples were prepared by dissolving the peptide to the concentration of ϳ1.5 M in 1, 20, 30, and 40 (v/v) TFE/H 2 O mixture, pH 4.0. In the experiments carried out in the presence of SDS micelles, the SDS concentration was increased by addition of the required volume of a stock solution (400 M SDS at pH 4.0). CD measurements were carried out in the spectral range of 190 -260 nm. Four consecutive scans per sample were performed in a 1-mm cell at 20°C. After subtraction of the CD signal of the solvent and correction for dilution, in the case of SDS addition, the observed ellipticity, (millidegree) was converted to the mean residue molar ellipticity [] (deg⅐cm 2 ⅐dmol Ϫ1 ).
NMR Spectroscopy-The sample for NMR measurements was prepared by dissolving HSP1 in 600 l of a TFE/H 2 O (30:70) (v/v) mixture to yield a peptide concentration of ϳ3.0 mM, pH 4.0. All two-dimensional 1 H NMR experiments were carried out in a Varian Unity 500 spectrometer operating at 499.730 MHz for 1 H frequency at a temperature of 20°C. The proton chemical shifts were referenced to 4,4dimethyl-4-silapentane-1-sulfonate (0.00 ppm). To assign the peptide resonance peaks, standard methods were used including Double Quantum-filtered (DQF)-COSY (24), total correlation spectroscopy (TOCSY) (25), nuclear Overhauser enhancement spectroscopy (NOESY) (26,27), and rotating frame Overhauser spectroscopy (ROESY) (28) experiments. The TOCSY spectra were acquired using a DIPSI spin-lock sequence at a field strength of 10 kHz and spin-lock evolution time of 70 ms. NOESY spectra with a 200-, 300-, and 400-ms mixing time were recorded to evaluate the spin diffusion. ROESY experiments were recorded with 100 ms.
All two-dimensional experiments were acquired in the phase-sensitive mode using the method of States et al. (29). The spectra width was typically 6000 Hz, and 512 1 increments, with 32 transients of 2048 complex points for each free induction decay, were recorded, except for DQF-COSY. The DQF-COSY experiment was collected with 512 1 experiments with 32 transients of 8192 complex points. Water suppression was achieved by low power continuous wave irradiation during the relaxation delay. Solvent suppression in ROESY experiments was carried out using the WATERGATE method (30). Data were processed on a Silicon Graphics Octane work station using the nmrPIPE/nmrVIEW (31).
Prior to Fourier transformation, the time domain data were zerofilled in both dimensions to yield a 4 ϫ 2K data matrix. When necessary, a fifth-order polynomial base-line correction was applied after transformation and phasing. The DQF-COSY experiment was processed to 16,384 ϫ 2048 data matrix to get a maximum digital resolution for coupling constant measurements. To obtain distance constraints, cross-peak volumes were estimated from the 300-ms NOESY spectra and 100-ms ROESY spectra. To relate them to inter-proton distances, a calibration was made using the distance of 1.8 Å for the well defined geminal ␤-protons, and the NOE volumes were classified as strong, medium, and weak, corresponding to the upper bound constraints of 2.8, 3.5, and 5.0 Å, respectively. Lower bounds were taken to be the sum of the van der Waals radii (1.8 Å) for the interacting protons in all cases.
Molecular Modeling-The three-dimensional structure of HSP1 was computed using the simulated annealing methods in the DYANA refine module (32). The modeling protocol was based on the methods implemented by Pristovšek et al. (33). Each round of refinement started with 20 random conformers, from which the 10 with the lowest target function were used to analyze constraint violations and assign additional NOE constraints to be included in the subsequent calculation. This process was repeated until all violations were eliminated. In the final round of refinement, a total of 100 structures was calculated, and the 40 conformers with the lowest target function were considered for analysis. After simulated annealing, these 40 structures with the target function smaller than 0.44 Ϯ 0.01 Å 2 , with no distance violation larger than 0.2 Å and no dihedral angle violation greater than 5°, were energy-minimized with full consistent valence force field (34, 35) (Morse and Lennard-Jones potentials, coulombic term). In the early stages of refinement, steepest descents were used until the maximum derivative was less than 1.000 kcal/Å. In the final round, we used conjugated gradients until the maximum derivative was less than 0.001 kcal/Å. All calculations were carried out on a Silicon Graphics Octane work station using the DISCOVER (Accelrys Inc., San Diego) software package, together with the INSIGHT II as graphic interface. The quality of the final structures was analyzed using the PROCHECK-NMR (36) and CNS (37) programs.
Atomic Force Microscopy-E. coli ATCC 25922 cells in mid-logarithmic phase were prepared as described under "Antibacterial Assays." E. coli (10 8 cells/ml) in trypticase soy broth were incubated with HSP1 (24.4 M) at 37°C for 4 h. After incubation, cells were collected and washed in 0.1 M sodium cacodylate buffer, pH 7.4, following centrifugation at 300 ϫ g for 5 min. Controls were processed in the absence of peptide. The cells were fixed by suspending them in Karnovsky's solution (2% paraformaldehyde, 2% glutaraldehyde, and 0.1 M sodium cacodylate buffer, pH 7.4). After fixation for 30 min, the samples were centrifuged at 300 ϫ g for 5 min, and the pellet was suspended in 20 l of 0.1 M sodium cacodylate buffer. The suspension was placed onto freshly cleaved mica and air-dried. The mica was fixed on the specimen holder with a two-sided adhesive tape and was then installed on the top of the AFM support for observation. The AFM used in this experiment was a TopoMetrix 2000 Explorer (TopoMetrix) operating in the contact mode and at ambient air. A piezoelectric hybrid tube scanner with a maximum scanning area of 2.3 m square and a standard 200-m V-shaped Si 3 N 4 cantilever with integrated pyramidal tips was used. The nominal spring constant for the contact force of the tip on the specimen surface was set to 0.00 nA. The line scan speed was set to 4 m/s. different components. All collected fractions were lyophilized and submitted to mass spectrometry. The fraction HPA43, containing the peptide of interest, was named Hylaseptin P1 (Fig. 1a).

Isolation and
Identification and Characterization-The primary structure was determined by de novo sequencing and confirmed by automatic Edman degradation in a single analysis for both cases (Fig. 2 and Table I). The resulting sequence was GILDAIKA-IAKAAG-COOH, and its calculated molecular mass was 1311.80 Da with an estimated pI of 8.59. All leucine and isoleucine v and w ions could be observed and assigned in the HSP1 sequence, in accordance with the result of Edman degradation. A search of the Swiss-Prot data bank showed 64.3% identity with the N-terminal portion of the brevinin-2 peptides from the Rana genus.
Chemical Synthesis-Solid phase Fmoc synthesis yielded 14 a MIC, minimal peptide concentrations required for total inhibition of cell growth. In liquid medium. These analyses were performed according to the recommendations of NCCLS. Experiments were performed in triplicate. b NDA, no detectable activity.

FIG. 2. De novo sequencing of the fraction HSP1 [M ؉ H] ‫؍‬ 1311.80 in an ABI 4700 proteomics analyzer with TOF-TOF optics using N 2 as collision-induced dissociation gas.
The observed fragments allowed complete assignment of the major y and b series of ions. The peptide sequence using the one-letter code following the y series orientation is shown on the top part of the graph. mg of peptide, which was further purified by preparative and analytical RP-HPLC, using a three-step gradient of H 2 O/acetonitrile, and the native peptide was run as a blank (Fig. 1b). The purified synthetic HSP1 was also sequenced by MS/MS in order to confirm its amino acid sequence and homogeneity.
Biological Activities-The peptide HSP1 was tested against Gram-positive and Gram-negative bacteria strains, pathogenic for mammalian species. The Gram-positive strain was S. aureus ATCC 25923, and the two Gram-negative strains were P. aeruginosa ATCC 27853 and E. coli ATCC 25992, obtained from the SABIN Laboratory culture collection. The HSP1 MICs were compared with those for commercial antibiotics, under the same experimental conditions. The results showed that  (Table II).
In addition, the MICs of HSP1 were measured for the inhibition of growth of some pathogenic Candida spp. yeasts. In this case, the MFCs were also determined (Table III). Two Candida strains, C. albicans ATCC 1023 and C. albicans, isolated from clinical patients and resistant to amphotericin B were tested with HSP1 and some commercial fungicidal drugs, at the same molar ratio. The HSP1 MICs against both strains were lower than those determined for fluconazole, of which the MFC could not be determined because this drug is known for its fungistatic activity only.
As for ketoconazole and nystatin Table III shows that the MICs and MFCs are much lower than those observed for HSP1. Nevertheless, in this case to make a meaningful comparison, it is necessary to take into consideration that they are toxic drugs. The most common adverse effect of ketoconazole is gastrointestinal intolerance, followed by fatigue, abnormalities of liver function, and skin changes, besides a variety of other, less frequently observed, adverse effects (38 -40). In the case of nystatin, its administration is limited by dose-dependent toxicities such as renal toxicity (41). Therefore, for these reasons any compound that could substitute for them should be considered of high pharmacological relevance, and this fact makes HSP1 an interesting option.
The hemolytic activity of HSP1 was checked by incubating the peptide with erythrocytes in saline (Fig. 3). The same maximal concentration of 256 g/ml (195.2 M) used to test HSP1 against the growth of the various microorganisms produced a minimal hemolytic effect (inferior to 5%). Noteworthy, this concentration is 32 times higher than the measured MIC for HSP1 against S. aureus.
The cytometric analyses of the whole blood also showed that HSP1 does not produce any detectable damage in white cells  ( Fig. 4, a and b). In fact, even after 4 h of incubation with HSP1, at the same concentrations used to kill bacteria and fungi, the number of all blood cellular types did not show any significant alteration. Similarly, there was no change in the mean platelet volume. CD- Fig. 5a reports the CD spectra of HSP1 in water and in the presence of TFE. In water the peptide is in a random conformation as indicated by the negative band at about 200 nm. The addition of TFE induces a conformational transition to an ␣-helical secondary structure that reaches a plateau above 20% TFE (Fig. 5a, inset). A similar behavior is observed when the peptide is in the presence of SDS micelles (Fig. 5b). In this case, only 4 mM SDS was sufficient to favor the transition of HSP1 to a helical conformation.
NMR Structure-Because the CD spectra indicate that the peptide acquires an ␣-helical conformation in both TFE and SDS, its solution structure was determined in a 30:70 (v/v) TFE/H 2 O mixture. Sequential resonance assignments were performed based on the combined use of two-dimensional TOCSY and NOESY (42). 1 H chemical shifts and 3 J NH ␣ coupling constants for the peptide were deposited in the BioMagResBank (BMRB) data base under accession number 5890. Fig. 6 shows the plot of the residual ␣H chemical shifts. The negative values observed from residues 2 to 13 suggest the presence of an ␣-helix segment in that region. The summary of the inter-residue NOEs for the peptide is reported in Fig. 7. The strong d NN (i,i ϩ 1) NOE together with a d␣ N (i,i ϩ 3), d␣␤(i,i ϩ 3), and d␣ N (i,i ϩ 4) and a weaker d␣ N (i,i ϩ 1) clearly indicate the peptide is in an ␣-helical configuration, consistent with the indication of the residual chemical shift and with the CD data.
Molecular Modeling-Out of 100 structures that were generated, 40 structures based on the minimum target function (0.44 Ϯ 0.01 Å 2 ) were energy-minimized using the software DISCOVER. After minimization, 24 structures based on the low r.m.s.d. were used to represent the solution three-dimensional structure of HSP1. The pairwise r.m.s.d. of the 24 structures, measured over the all peptide sequence, is 0.04 Ϯ 0.03 Å for the backbone and 0.25 Ϯ 0.16 Å for the heavy atoms. Table   FIG  IV shows the structural statistics of the selected 24 solution structures of HSP1.
As judged by the low r.m.s.d. values from idealized geometry for bond and angle, the structures exhibit a very good covalent geometry and stereochemistry. The Ramachandran plot confirms the high quality of the 24 structures of HSP1, which shows 100% of and angles in the most favored region. Fig.  8 displays the superposition of the 24 final models of HSP1 and, confirming the high definition of the three-dimensional structures, showing that the peptide can fold as an amphiphilic ␣-helix.
AFM-Atomic force microscopy was employed to examine the effect of HSP1 on the E. coli ATCC 25922 morphology of the cells. The observation of a roughness on the bacterial surface when using the peptide at MIC concentration (Fig. 10b), which was not observable in the untreated sample (Fig. 10a), might depict the initial stages of peptide-caused permeability of the cell wall of E. coli. These results support the hypothesis that the inhibition of bacterial growth is associated with the perturbation of the cell wall membrane. DISCUSSION The dermal glands of frogs produce antimicrobial peptides that protect the animal against noxious microorganisms and assist them in wound repair. The amino acid sequences of these peptides are very dissimilar, both within and between species, so that the 5,000 living anuran frogs may produce approximately 100,000 different antimicrobial peptides. The presence of antimicrobial peptides with such variability in sequence and a wide spectrum of action represents the success of evolution in providing frogs with maximum protection against infectious microbes and, at the same time, minimizing the chance of microorganisms developing resistance to individual peptides. In other words this is the dream of drug designers (43).
A sequence search in the Swiss-Prot Data Bank for the peptide HSP1 revealed more than 64% identity with the Nterminal fragment of the antimicrobial lytic peptides known as brevinin-2 from the skin secretion of the Rana genus (10). This family of antimicrobial peptides displays a highly conserved pattern of disulfide bonds, forming a ring with the seven Cterminal residues that are absent in HSP1. Despite the numerous attempts to find S-S bond containing peptides in crude skin extracts of H. punctata, it was not possible to identify any brevinin-like peptides. The hypothesis that HSP1 could be a truncated form of brevinin-2 present in the genome of H. punctata is under investigation. The lack of a cyclic motif, however, does not affect the antimicrobial properties of HSP1 that, in fact, shows the ability to kill both Gram-positive and Gramnegative bacteria strains such as S. aureus, P. aeruginosa, and E. coli similarly to brevinin-2, with the advantage of being smaller and easier to synthesize or overexpress for biotechnological use.
An essential requirement for any antimicrobial host defense or therapeutic agent is that it should affect specific microbial targets (44). Cationic peptides exhibit a high affinity for lipopolysaccharides (LPS), which are the main components of the outer membrane of bacteria. These peptides completely displace the LPS-bound divalent cations that stabilize the structure of the membrane (45). The LPS binding capacity of antimicrobial peptides is of great clinical advantage compared with classic antibiotics, because it prevents endotoxemia (46). Massive bacterial lysis induced by antibiotic intake generates an uncontrolled systemic LPS release, which in turn promotes proinflammatory cytokine production and finally leads to septic shock, often with fatal consequences. The outer membrane of higher eukaryotes is made of electrically neutral phospholipids like phosphatidylcholine and sphingomyelin, whereas bacterial membranes have exposed negatively charged phosphatidylglycerol and cardiolipin (47). Another difference is the lack of cholesterol in bacterial membranes. These parameters were found to be important for the selectivity of antimicrobial peptides (48). Indeed, our data show that HSP1 is a candidate for a new antimicrobial drug. In fact, the peptide is able to inhibit, at low concentrations, some pathogenic bacteria (MIC 6.1 M for S. aureus and 24.4 M for E. coli). In this respect, it is worthwhile to point out that we have tested the peptide against an inoculum of 10 7 colony-forming units/ml, 1000 times more concentrated than normally used (10 6 colony-forming units/ml) (8,49,50).
As both humans and fungi are eukaryotic organisms, antifungal agents may affect the cellular metabolism of both the host and the pathogen. Thus, relatively few antifungal agents with minimal toxicity and few side effects are available compared with the plethora of antibacterial drugs. In the past 2 decades, systemic fungal infections, essentially invasive candi-diasis, have increased substantially, and they are considered an important cause of morbidity and mortality in cancer patients (51). Azole (imidazole and triazole) compounds such as ketoconazole, fluconazole, itraconazole, and polyenes, such as nystatin and amphotericin B, AmB, are among the most potent anti-fungal agents (52). Ketoconazole is used worldwide as an oral anti-fungal agent with a broad spectrum of action against both superficial and systemic mycosis (53). AmB remains the drug of choice for treatment of most fungal diseases because of its broad spectrum and potent fungicidal activity, although significant side effects, such as renal toxicity (41), limit its clinical utility. The azole antifungal agents are easier to take and less toxic than AmB, but their use is limited by multiazoleresistant strains (54). As a result, despite such a range of available antifungal drugs, these infections are, as yet, associated with significant morbidity and mortality. In the antimycotic tests reported here, we have shown that HSP1 displays an apparently unattractive MFC for C. albicans growth inhibition (MFCs 195.2 M for both wild type and ATCC 1023). However, a meaningful comparison of these data with the data for the currently used antifungal drugs must take into account the fungicidal properties of HSP1 and not only the fungistatic activity, i.e. the major feature for fluconazole, for example. Moreover, HSP1 shows very low hemolytic activity, only 4.7% at 256 g/ml (195.2 M), and causes no detectable damage to white blood cells. Thus it appears to be a compound with low toxicity and consequently should generate very few risks for humans. Therefore, overall the pharmacological properties of HSP1 render it a quite promising and attractive potential antifungal agent.
In devising new and more potent antimicrobial peptides, one important aspect is the knowledge of the mechanisms by which they cause cell death. A long held paradigm for microbicidal action is that peptides kill the microorganisms by causing multiple and insurmountable perturbations in the microbial cell membranes (44). The structural features of HSP1 and its antimicrobial and antifungal properties correlate perfectly well with this theory.
The CD spectra indicate that HSP1 does not display any regular secondary structure in aqueous solution. However, in the presence of TFE or of SDS micelles, mimetic of a membrane environment, the CD profile clearly indicates that it acquires an ␣-helical conformation. In addition, the NMR-derived threedimensional structure shows that the ␣-helix has an amphiphi- lic nature. In fact, Fig. 9 clearly shows the presence of a well defined charged surface region, mostly positive, localized at the peptide C terminus, whereas a hydrophobic region, on the opposite peptide surface, is encountered at the N terminus. This conformational transition from random coil to an amphiphilic ␣-helix when moving into membrane-mimetic environments and the net separation of the polar and hydrophobic regions reflects its potential to interact efficiently with biomembranes.
AFM has been used extensively to study biological samples and is considered a useful methodology to investigate a number of cell membrane features (2). In Fig. 10, the bubble-like membrane formations on the surface of bacteria treated with HSP1, contrasting with the smoother surface of the untreated cells, strongly supports the hypothesis that the ability of HSP1 to kill cell pathogens is dependent on its direct interaction with the cell wall membrane, leading to a progressive process of permeabilization and culminating in cell lysis. Overall, these results are in accordance with the current paradigm that antimicrobial activity of peptides is associated with their capability to destabilize the membrane structures of pathogens (6,7,45).

CONCLUSIONS
In the past decade, the function of cationic antimicrobial peptides has become increasingly apparent, and there is a growing body of evidence that their role in defense against microbes is as important to the host as antibodies, immune cells, and phagocytes (5). The present investigation of the cationic antimicrobial peptide HSP1, obtained from an overlooked frog species widespread in the Brazilian rain forest, is an example of how important is the study of natural compounds and how crucial is the preservation of biodiversity. Investigations are now in progress to better understand the organization of the peptide on the pathogen cell wall and to optimize its pharmacological specificity.