M. L. Mangoni, Unità di Diagnostica Molecolare Avanzata, II Facoltà di Medicina e Chirurgia, Azienda Ospedaliera S. Andrea, via di Grottarossa, 1035-00189 Roma, Italy Fax: +39 06 33776664 Tel: +39 06 33775457 E-mail: email@example.com
Antimicrobial peptides constitute one of the main classes of molecular weapons deployed by the innate immune system of all multicellular organisms to resist microbial invasion. A good proportion of all antimicrobial peptides currently known, numbering hundreds of molecules, have been isolated from frog skin. Nevertheless, very little is known about the effect(s) and the mode(s) of action of amphibian antimicrobial peptides on intact bacteria, especially when they are used at subinhibitory concentrations and under conditions closer to those encountered in vivo. Here we show that esculentin-1b(1–18) [Esc(1–18)] (GIFSKLAGKKLKNLLISG-NH2), a linear peptide encompassing the first 18 residues of the full-length esculentin-1b, rapidly kills Escherichia coli at the minimal inhibitory concentration. The lethal event is concomitant with the permeation of the outer and inner bacterial membranes. This is in contrast to what is found for many host defense peptides, which do not destabilize membranes at their minimal inhibitory concentrations. Importantly, proteomic analysis revealed that Esc(1–18) has a limited ability to modify the bacterium’s protein expression profile, at either bactericidal or sublethal concentrations. To the best of our knowledge, this is the first report on the effects of an antimicrobial peptide from frog skin on the proteome of its bacterial target, and underscores the fact that the bacterial membrane is the major target for the killing mechanism of Esc(1–18), rather than intracellular processes.
fluorescein isothiocyanate–dextran of 4 kDa average molecular mass
fluorescein isothiocyanate–dextran of 10 kDa average molecular mass
fluorescein isothiocyanate–dextran of 40 kDa average molecular mass
fluorescein isothiocyanate–dextran of 70 kDa average molecular mass
large unilamellar vesicle
minimal inhibitory concentration
outer membrane protein
peptide mass fingerprinting
scanning electron microscopy
transmission electron microscopy
Numerous families of ribosomally synthesized antimicrobial peptides, from virtually all life forms, have been described [1,2]. They are conserved components of the innate immune system in plants and animals, and represent the most ancient and efficient weapon against microbial pathogens . In recent years, for several antimicrobial peptides, additional chemokine-like and immunomodulatory activities have been reported; these are involved in infection processes leading to the appropriate activation of adaptive immune responses in higher vertebrates . For this reason, these molecules are more properly referred to as host defense peptides .
An increasing number of microorganisms have become resistant to a multiplicity of clinically used drugs, causing a severe crisis in the treatment and management of infectious diseases, with serious consequences for human health . Therefore, substantial efforts have been devoted to identifying new classes of antibiotics displaying diverse mode(s) of action: antimicrobial peptides are currently considered to be some of the most promising candidates for the development of novel anti-infective preparations [7,8]. Although antimicrobial peptides show marked variation in size, sequence, and conformation, most of them are polycationic, and fold into an amphipathic helical or β-sheet structure .
Numerous articles have provided compelling evidence that many antimicrobial peptides penetrate microbes and interfere with general intracellular functions (e.g. DNA, protein and cell wall synthesis or chaperone-assisted protein folding) without destabilizing their plasma membrane. Some examples are as follows: (a) buforin 2, from histone H2A of Bufo bufo, and PR-39, from pig intestine ; (b) derivatives of pleurocidin, a fish-derived antimicrobial peptide, and dermaseptin, from frog skin ; (c) drosocin and pyrrhocoricin, from insects ; and (d) Bac-7(1–35), corresponding to the 35-residue N-terminal region of Bac-7 from bovine neutrophils .
However, very little is known about the effect(s) of antimicrobial peptides at subinhibitory concentrations. Also, as reported in the literature, the antibacterial activities of a vast repertoire of host defense peptides have been assayed only in buffered or dilute media, and these peptides have been found to be ineffective in the presence of physiological ionic strength or biological fluids such as serum . Hence, intense research focusing on antimicrobial peptides is currently directed at completing our knowledge of their mode(s) of action at both lethal and sublethal doses and at shedding light on their antimicrobial properties under physiological conditions.
Among the natural sources for antimicrobial peptides, the granular glands of amphibian skin constitute one of the richest [14–16]. Studies on the mode of action of amphibian antimicrobial peptides have mainly addressed their interaction with phospholipid bilayers, but some have also dealt with intact microbes, and revealed that these antimicrobial peptides can perturb both model and biological membranes [17–19]. We have recently compared the killing activities of antimicrobial peptides belonging to families that include esculentins, temporins, and bombinins H, extracted from three different species of anurans, against multidrug-resistant clinical isolates . These studies showed that esculentin-1b(1–18) [Esc(1–18), GIFSKLAGKKLKNLLISG-NH2], the amidated form of a linear peptide encompassing the first 18 residues of the full-length esculentin-1b (46 amino acids) from the skin of Pelophylax lessonae/ridibundus (previously classified as Rana esculenta ), was the most potent peptide, particularly towards Gram-negative species, with a minimal bactericidal concentration ranging from 0.5 to 1 μm, in sodium phosphate buffer .
Here, to expand our knowledge of the activity of Esc(1–18) against Gram-negative bacteria, along with the underlying molecular mechanism, we analyzed the effect(s) of this peptide on Escherichia coli ATCC 25922 by investigating the following: (a) its microbicidal action and kinetics in different media; (b) its ability to permeate both artificial and bacterial membranes; (c) its affinity of binding to lipopolysaccharide (LPS); (d) its ability to synergize with conventional antibiotics; and (e) its effects on bacterial morphology and the bacterial proteome.
Our data have shown that this unique amphibian-derived peptide: (a) kills E. coli via membrane perturbation; (b) strongly synergizes with erythromycin, presumably by increasing the intracellular influx of this drug, as a result of increased membrane permeability; (c) elicits identical changes in the bacterium’s protein expression pattern at lethal and sublethal concentrations; and (d) preserves antibacterial activity under conditions closer to those encountered in vivo. This is in contrast to many other host defense peptides, which kill microorganisms by altering intracellular processes, and become inactive in physiological solutions. Importantly, to the best of our knowledge, this is the first demonstration of how an amphibian antimicrobial peptide can affect the protein expression profile of its bacterial target.
The secondary structure of Esc(1–18) was determined by using CD spectroscopy in 10 mm sodium phosphate buffer (pH 7.4) and when bound to phosphatidylethanolamine (PE)/phosphatidylglycerol (PG) vesicles of composition 7 : 3 (w/w), which is typical of the E. coli inner membrane (IM) . As indicated in Fig. 1A, the peptide conformation in buffer was predominantly disordered, whereas association of the peptide with lipid vesicles induced a transition to a predominantly α-helical conformation. Complete binding of the peptide to the lipid vesicles was manifested by the absence of significant changes in the CD spectrum when the lipid/peptide molar ratio was increased from 100 to 400. The helical wheel diagram of Esc(1–18) in a perfect α-helical conformation (Fig. 1B) shows amphipathicity of the peptide, with hydrophobic and hydrophilic residues segregating on opposite sides of the molecule.
The activity of Esc(1–18) against E. coli ATCC 25922 was first evaluated by the microdilution broth assay to determine the minimal inhibitory concentration (MIC), using both a standard inoculum of 1 × 106 colony-forming units (CFUs)·mL−1 and 4 × 107 CFU·mL−1, as most of the experiments described below needed this higher number of bacterial cells. As shown in Table 1, where the frog skin membrane-active peptide temporin-1Tl  is included as a reference, the MIC of Esc(1–18) in culture medium (Mueller–Hinton broth) was found to be directly correlated with the number of microbes present in the inoculum. Afterwards, to examine the killing activity of Esc(1–18) against E. coli and to determine whether this process was affected by the ionic strength of the incubation medium, we assayed the peptide’s bactericidal effect, as defined in Experimental procedures, after 1.5 h of incubation with bacteria, either in Mueller–Hinton broth, sodium phosphate buffer (pH 7.4), or NaCl/Pi (Table 1). Interestingly, in all cases, a reduction in the number of viable cells of ≥ 3 log10 CFU·mL−1 (99.9% mortality) was achieved at twice the MIC (16 μm) when a standard inoculum was used. In contrast, with the higher number of bacteria (4 × 107 CFU·mL−1), Esc(1–18) displayed a bactericidal effect at 32 μm, a concentration equal to the MIC, under these conditions (Table 1). Furthermore, to estimate the peptide’s ability to retain such activity under experimental conditions closer to those encountered in vivo, antimicrobial assays were performed in the presence of human serum. It is noteworthy that, unlike temporin-1Tl (Table 1) and other natural antimicrobial peptides, such as human β-defensin 2 and dermaseptin S, which lost their bacteriostatic effect in the presence of 20–30% serum (MIC ≥ 200 μm) [24,25], Esc(1–18) was able to partially preserve its antibacterial activity at a higher serum percentage (70%), with MIC and bactericidal concentration values of 32 and 64 μm, respectively (Table 1), using a standard inoculum. As the peptide’s degradation by serum enzymes was prevented by heating serum at 56 °C (see Experimental procedures), our findings suggest that serum components do not strongly bind to Esc(1–18) and therefore do not significantly affect the availability of active peptide molecules.
Table 1. Antibacterial activity of Esc(1–18) and temporin-1Tl on E. coli ATCC 25922. The bactericidal activity is defined as the concentration of peptide that is sufficient to reduce the number of viable bacteria by ≥ 3 log10 CFUs·mL−1 after 1.5 h of incubation. The values found for temporin-1Tl are in parentheses.
Bactericidal activity (μm)
Sodium phosphate buffer (pH 7.4)
1 × 106
32 (> 128)
64 (> 128)
4 × 107
The killing kinetics occurred on a quite fast time scale, causing more than 90% microbial deaths within 10 min, at the MIC (Fig. 2). The latter result indicates a substantial difference from those antimicrobial peptides that preferentially act on intracellular targets and over a longer time scale, and do not manifest any lethal activity at their MICs [11,26].
Mode of action studies
It is well known that the mode of action of antimicrobial peptides depends on the mode(s) of their interaction with the cell membrane. However, before reaching it, the peptide needs to bind and traverse the cell wall, which, in Gram-negative bacteria, is surrounded by an outer membrane (OM), composed mainly of the anionic LPS (or endotoxin), which forms a barrier to protect bacteria from many hydrophilic and hydrophobic molecules, including some antimicrobial peptides . Therefore, we first investigated the ability of Esc(1–18) to bind LPS and penetrate the E. coli OM.
LPS binding properties
LPS films have been used as suitable model systems to mimic the outer layer of the Gram-negative OM [28,29]. To investigate the binding of Esc(1–18) to LPS, we monitored the changes in surface pressure of monolayers of commercially available LPS from E. coli O111:B4 upon a peptide’s insertion, using the method described in Experimental procedures. Esc(1–18) efficiently penetrated E. coli LPS monolayers, as manifested by the increase in film surface pressure (Fig. 3). Under experimental conditions, monolayer penetration reached a substantial stability around 1.0 μm Esc(1–18) (Fig. 3A), which was then selected as the peptide concentration for subsequent experiments. When data from similar measurements were analyzed in terms of change in surface pressure (Δπ) versus initial surface pressure (π0), the critical surface pressure corresponding to the LPS lateral packing density preventing the intercalation of Esc(1–18) into E. coli LPS films could be derived by extrapolating the Δπ−π0 slope to Δπ = 0, yielding a value of ∼ 47 mN·m−1 (Fig. 3B). The kinetics of the insertion of the peptide into the LPS monolayer were characterized by a rapid and marked enhancement of surface pressure that followed soon after injection of the peptide into the subphase, the lag phase for this process being too short to be measurable with our instrumentation (Fig. 3C). In a typical experiment, within the first 60 s after peptide injection, π attained a value that was slightly over 85% of the value recorded at the end of measurement (Fig. 3C). This initial surge was then followed by a slower increase in π for approximately the next 19 min, when a plateau was reached, and no more significant variation in π was observed for at least the next 16 min. This general kinetics pattern was apparently independent of the initial surface pressure and from peptide concentration, and was similar to that recorded for temporin-1Tl interacting with a monolayer made of the same type of LPS .
The permeabilization of the OM was determined by investigating the periplasmic β-lactamase activity against its specific substrate CENTA . A plot of enzyme release, as a function of peptide concentration, is shown in Fig. 4A. Interestingly, there was a dose-dependent perturbation of the OM, and the greatest perturbation was obtained at the MIC of the peptide (32 μm with 4 × 107 CFU·mL−1). The rate of CENTA hydrolysis, upon addition of 1 × MIC of Esc(1–18) to the cells, was also monitored for 20 min, and the amount hydrolyzed was found to be ∼ 70% of the total within the first 5 min (Fig. 4B).
Next, the effect of the peptide on the E. coli IM was analyzed by measuring the intracellular influx of SYTOX Green . This cationic dye, which is excluded by intact membranes, but not from those with lesions large enough to allow its entrance, dramatically increases its fluorescence when bound to intracellular nucleic acids (Fig. 5). The data revealed that Esc(1–18) augmented the permeability of the IM, with kinetics superimposable on those of the OM permeation (although with a slightly longer lag time), reaching a final value in about 15–20 min and in a concentration-dependent fashion. However, membrane permeation caused by Esc(1–18) was not maximal at levels up to twice the MIC. This was manifested by a further enhancement of fluorescence, following the addition of a detergent for the complete solubilization of phospholipid bilayers (Fig. 5, arrow at 20 min). Then, to investigate the size of membrane lesions induced by the peptide, we assessed the release of intracellular compounds, such as the cytoplasmic β-galactosidase, whose Stokes radius is equal to 69 Å . As reported in Fig. 6, the enzyme release was almost 40% of maximum when the peptide concentration was equal to the MIC. These results underscore a disturbance of the IM, although to a smaller extent than that of the OM, and indicate the existence of a direct correlation between the peptide dose and the extent of both microbial death and membrane disturbance.
Synergistic activities with conventional antibiotics
Checkerboard titrations were carried out using Esc(1–18) in combination with different classes of clinically available antibiotics. As illustrated in Table 2, a clear synergism was noted when the peptide was mixed with cephalosporin C, erythromycin, nalidixic acid, netilmicin, and rifampicin [a fractional inhibitory concentration (FIC) ≤ 0.5 indicates synergy; see Experimental procedures]. To obtain insights into the mode of action underlying the synergistic activity, we investigated the bactericidal action of the combination of Esc(1–18) and erythromycin, the antibiotic that gives the best synergy with the peptide, as indicated by the lowest FIC (Table 2). Erythromycin is a hydrophobic molecule that inhibits protein synthesis by blocking either the peptidyltransferase reaction or the translocation step, but cannot easily traverse the OM of Gram-negative bacteria .
Table 2. Interaction of Esc(1–18) with conventional antibiotics against E. coli ATCC 25922. The ranges of concentrations tested were as follows: 0.25–64 mg·L−1 for Esc(1–18) and 0.25–256 mg·L−1 for the other antimicrobial agents. FIC indices were interpreted as follows: FIC ≤ 0.5, synergy; 0.5 < FIC <1, additivity; 1 ≤ FIC < 4, indifference; and FIC ≥ 4, antagonism.
As expected, erythromycin displayed a weak bactericidal effect, causing about 35% microbial death at a very high concentration (256 μg·mL−1) and within 3 h of incubation (Fig. 7). Interestingly, when sublethal concentrations of Esc(1–18) and erythromycin were combined, ∼ 8% and 90% killing were detected after 20 min and 3 h, respectively (Fig. 7). These results provide additional support for the membrane-permeabilizing properties of Esc(1–18). Indeed, as no reduction in the number of viable cells was observed within the first 20 min [killing kinetics of Esc(1–18)], but a reduction was observed after a longer time (2–3 h) (Fig. 7), corresponding to the time-kill kinetics of erythromycin, we can assume that the synergistic activity between the two compounds is the result of increased access of erythromycin to its intracellular target, because of increased peptide-induced permeability of the cytoplasmic membrane and/or the LPS layer.
Permeabilization of large unilamellar vesicles (LUVs)
The peptide’s ability to alter the structure of the plasma membrane of E. coli cells by a nonstereospecific process was also confirmed by employing calcein-loaded liposomes made of PE/PG (7 : 3, w:w). Different concentrations of peptide were added to LUV suspensions, and membrane permeability was measured by following fluorescence recovery due to calcein leakage from the liposomes . Calcein leakage occurred immediately after peptide addition, and reached a plateau within the first 15 min (Fig. 8A). Figure 8B shows the dose–response curve of peptide-induced calcein release from PE/PG vesicles. The data clearly show a membrane-perturbing activity of Esc(1–18). Note that this activity increased in a dose-dependent manner and reached its maximum (∼ 65% calcein leakage) at a peptide/lipid molar ratio of 1.5. These results are comparable with those found for other membrane-active antimicrobial peptides, such as cathelicidin LL-37 . However, at a peptide/lipid molar ratio as low as 0.04, Esc(1–18) was more active than cathelicidin LL-37 . As illustrated in Fig. 8B, Esc(1–18) did not fully permeabilize the lipid vesicles, and the calcein leakage diminished when the peptide/lipid molar ratio exceeded 1.5, probably because of the peptide’s aggregation at high concentrations. Taken together, these observations are in line with those made above using intact cells (Figs 5 and 6), and are consistent with the suggestion that Esc(1–18) binds and destabilizes the bacterial membrane, but to a lesser extent than temporin-1Tl . According to what has been stated for other antimicrobial peptides , such a discrepancy between the two frog skin peptides might be related to a higher fraction of membrane-bound active temporin-1Tl than of Esc(1–18).
The ability of Esc(1–18) to induce the leakage of liposome-encapsulated markers of different sizes was also monitored. PE/PG LUVs were preloaded with fluorescein isothiocyanate–dextrans (FITC-Ds) of 4, 10, 40 or 70 kDa average molecular mass (FITC-D 4, FITC-D 10, FITC-D 40, and FITC-D 70), and then incubated with the peptide. The data shown in Fig. 9 reveal that Esc(1–18) is able to cause the release of the four dextrans used in a dose-dependent manner, and with a dependence on the size of the liposome-entrapped probe. This indicates that Esc(1–18) does not have a detergent-like effect on the membrane , and that membrane lesions produced by this peptide are larger than 58 Å (Stokes radius of FITC-D 70 ), which is in agreement with its ability to promote the release of β-galactosidase from E. coli cells.
Scanning electron microscopy (SEM)
The effect(s) of Esc(1–18) on E. coli morphology were visualized by SEM (Fig. 10). The exposure of 4 × 107 cells·mL−1 at the corresponding MIC of Esc(1–18) resulted in an irregular rod form with a deep wrinkling of the cell surface (within 5 min). However, all of these changes became more pronounced after a longer incubation time (20 min). With reference to untreated cells, bacteria appeared flat, with a collapsed cell structure and surface corrugation similar to that induced by temporin-1Tl , but in a milder form.
Transmission electron microscopy (TEM)
TEM was then used to directly examine the damage to bacteria induced by the peptide. A local disturbance to the membrane was noted after the first 5 min of peptide treatment, and this was followed by more damage and loss of cellular integrity, with a partial discharge of the cellular contents, within 20 min (Fig. 11). These results correlate with the killing kinetics of the peptide, and show that the antibacterial activity of Esc(1–18) is concomitant with its membrane-perturbing activity.
To determine whether Esc(1–18) could evoke a cellular reaction by modifying, within 20 min, the expression levels of proteins under conditions where the peptide did not affect the viability of E. coli or reduced it by ∼ 40% (2 and 16 μm peptide, respectively; data not shown), the bacterial proteome was analyzed by means of 2D-PAGE and MS. This analysis revealed a similar pattern of responses to both sublethal and lethal peptide doses, consisting of only a few significant variations in protein expression (11 protein spots) as compared with untreated cells. The majority of these spots (Fig. 12) were identified by peptide mass fingerprinting, reported in Table 3. In particular, reductions in the expression levels of a number of OM proteins (OMPs), such as OMPc, nmpC, and OMP F, all of which form passive diffusion pores allowing the passage of small molecular weight hydrophilic materials , were detected in peptide-treated bacteria (Table 3), with stronger reductions being seen at 16 μm Esc(1–18). Otherwise, a slight increase in OMP W expression was found at 16 μm. Note that the function of this protein is not completely understood; however, recent data have suggested that it may be involved in the protection of bacteria against various forms of environmental stresses . Overexpression of trigger factor (TF) was also caused by both peptide concentrations. TF in E. coli is a ribosome-associated chaperone that initiates folding of newly synthesized proteins . The enhanced production of TF might contribute to more streamlined denovo protein folding, by shielding nascent polypeptides on the ribosome, and thereby shortening degradation or aggregation processes . In addition, as shown in Table 3, exposure of bacteria to Esc(1–18) gave rise to a drop in the level of the following enzymes: (a) glucosamine-fructose-6-phosphate aminotransferase, which catalyzes the formation of glucosamine 6-phosphate, a precursor of cell wall peptidoglycan synthesis ; and (b) the dihydrolipoyllysine-residue succinyltransferase component of 2-oxoglutarate dehydrogenase complex, which catalyzes the conversion of α-ketoglutarate into succinyl-CoA as part of the tricarboxylic acid cycle .
Table 3. Protein spots identified by PMF.
UniProt accession no.
No. of matching peptides
Sequence coverage (%)
Peptide concentration (μm)
a The mascot score represents the probability that the observed match is a random event. Protein scores greater than 61 are significant (P < 0.05). b This spot contains three different OMPs. c The three indicated UniProt accession numbers correspond to glucosamine-fructose-6-phosphate aminotransferase from different E. coli strains. This protein was found in spot 6703 and spot 6707.
Outer membrane protein C precursor
Outer membrane porin protein nmpC precursor
Outer membrane protein F
Outer membrane protein W
Dihydrolipoyllysine residue succinyltransferase component of 2-oxoglutarate dehydrogenase complex
The repertoire of antimicrobial peptides has dramatically increased during the past two decades, and > 800 antimicrobial peptides have been isolated from different plant and animal sources, with more than 400 isoforms being obtained from amphibian species. This article discusses the antibacterial activity and mode of action of the N-terminal region of esculentin-1b, an antimicrobial peptide from the skin of P. lessonae/ridibundus. As no activity against microorganisms had been previously observed with the 19–46 fragment of this peptide, possibly because of its low positive charge at neutral pH (+1 versus +5 for the whole molecule) , we analyzed the antibacterial activity of the 1–18 N-terminal portion of esculentin-1b. Surprisingly, this activity was found to be similar to that of the full-length natural peptide [48,49], whereas complementary insecticidal properties were ascribed to the 19–46 fragment . Recent experiments have underscored the fact that Esc(1–18) possesses a wide spectrum of antimicrobial activity against several species of Gram-positive bacteria, Gram-negative bacteria, Candida and multidrug-resistant nosocomial pathogens, without being hemolytic [20,48].
Regardless of the precise mode of action, the effect(s) of antimicrobial peptides in general depends upon their interaction with the microbial membrane [51,52]. In particular, the first step in this process is the electrostatic attraction between the cationic peptide and the negatively charged components of the cell envelope, such as the phosphate groups within the LPS molecules of the OM in Gram-negative bacteria or the lipoteichoic acids on the surface of Gram-positive bacteria. In the case of Gram-negative bacteria, antimicrobial peptides initially cross the LPS layer, in a self-promoted uptake process driven by hydrophobic interactions, and subsequently reach the IM . Nevertheless, studies performed with intact bacteria have shown that antimicrobial peptides, e.g. pleurocidin derivatives and buforin 2, do not disturb the membrane of E. coli at their minimal antimicrobial concentrations, but rather traverse it, accumulate intracellularly, and damage a variety of essential vital processes to mediate the lethal event, which occurs only at multiples of the MICs [7,11,26].
In this study, we have shown that Esc(1–18) displays rapid bactericidal activity, at the MIC, against E. coli (Fig. 2), concomitant with alteration of its inner and outer membranes (Figs 4–6). As shown by the biophysical and biochemical assays, this peptide strongly bound LPS and completely permeated the LPS OM (Figs 3 and 4). In addition, the intracellular influx of SYTOX Green (Fig. 5), the extracellular leakage of β-galactosidase (Fig. 6), calcein and dextran release from liposomes mimicking the E. coli IM (Figs 8 and 9) and electron microscopy images (Figs 10 and 11) suggest that Esc(1–18) is a membrane-active peptide which kills bacteria by, primarily, injuring their membranes. This interpretation is further supported by the small changes in the proteomic profiling of bacteria upon treatment with either sublethal or lethal peptide doses.
Unlike DNA microarray analysis, which has proven to be a successful tool for the monitoring of whole genome expression profiles at the mRNA level , proteomic analysis has been found to be very useful for comparing changes in the expression levels of many proteins, under antibiotic treatment or other environmental conditions . Importantly, this approach represents the most powerful method for providing a better understanding of complex biological processes, as well as post-translational modifications of proteins, which cannot be obtained from mRNA expression profiles . In peptide-treated bacteria, a decrease in the levels of OMPc, OMP F, and nmp proteins, which allow the passive diffusion of hydrophilic molecules across the OM, would represent a cellular reaction that compensates for the stress provoked upon contact with a membrane-active antimicrobial peptide. In line with this explanation is the greater production of TF and OMP W, at the highest peptide concentration used, to guarantee bacteria a more protected environment, which would be particularly important for increasing their viability. Furthermore, the exposure of bacteria to Esc(1–18) would cause a slowdown of metabolic activities, which is in agreement with the lower levels of glucosamine-fructose-6-phosphate and dihydrolipoyllysine-residue succinyltransferase component of 2-oxoglutarate dehydrogenase complex.
Esc(1–18) did not cause bacteria to disintegrate and did not form blebs on their surface but, rather, emptied the cells, causing the loss of cellular material through the peptide-induced membrane breakages, and substantial roughening of their surface. The peptide might bind to the membrane surface in a carpet-like arrangement, inserting into the polar phospholipid headgroups. This would generate an unfavorable tension, resulting in the formation of transient breakages with a size larger than 58 Å, leading to bacterial death [9,56].
In addition, as suggested by the synergistic bactericidal activity of Esc(1–18) when combined with erythromycin, an increased peptide-induced membrane permeability, at subinhibitory peptide concentrations, would make it easier for hydrophobic drugs to enter the cells and to induce their toxic effect.
This work provides four interesting findings. The first is the ability of Esc(1–18) to display fast bactericidal activity, at the MIC, under both standard and physiological conditions. The second is its ability to simultaneously kill E. coli and permeate, in a dose-dependent manner, its outer and inner membranes, but without causing cell lysis. The third is the ability to modify, within 20 min, the expression levels of a limited number of bacterial proteins, at either lethal or sublethal concentrations. These findings rule out the possibility that variations in the production of these proteins account for the killing process of Esc(1–18). Note that only a few studies on the effect(s) of antimicrobial peptides on the proteomes of microorganisms have been performed to date. Interestingly, proteomic and transcriptomic analysis of the yeast Saccharomyces cerevisiae, following exposure to a similar antimicrobial peptide [esculentin-1a(1–21)], had shown downregulation of enzymes of the lower glycolytic pathway as well as a decrease in actin level, resulting in dramatic changes in cell physiology . It is worthy of remark that both fragments of esculentin peptide were found to affect the integrity of the microbial plasma membrane and the synthesis of the microbial cell wall. To the best of our knowledge, this study represents the first example of the effects of an antimicrobial peptide from frog skin on the proteome of bacteria, and demonstrates that the bacterial membranes are the major targets of its mechanism of action. Fourth, Esc(1–18) synergizes with conventional antibiotics in the inhibition of microbial growth. All of these properties, together with potent activity against a broad spectrum of multidrug-resistant clinical isolates  and a lack of lytic effects on human erythrocytes , lymphocytes, and keratinocytes (data not shown), make Esc(1–18) a very attractive membrane-active antimicrobial peptide for in-depth analysis of biological properties. More specifically, it can be considered to be a promising template for: (a) the production of less toxic anti-infective preparations with new modes of action and with the ability to elicit few changes in the proteome of the target microorganism and no microbial resistance; and (b) the design of potential coadjuvants of those antimicrobial agents that are already available.
Synthetic Esc(1–18) was purchased from GENEPEP (Prades le Lez, France). The purity of the peptide, its sequence and its concentration were determined as previously described . Culture media, antibiotics, 2-nitrophenyl β-d-galactoside (Gal-ONp) calcein and LPS from E. coli serotype O111:B4 were all purchased from Sigma (St Louis, MO, USA). SYTOX Green was from Molecular Probes (Invitrogen, Carlsbad, CA, USA). Egg yolk PG and PE were purchased from Avanti Polar Lipids (Alabaster, AL, USA). FITC-Ds were purchased from Sigma. All other chemicals were reagent grade. For antimicrobial assays, the commercially available quality control strain E. coli ATCC 25922 was used.
Penetration into LPS monolayers
Insertion of Esc(1–18) into LPS monolayers spread at an air/buffer (5 mm Hepes, pH 7) interface was monitored by measuring surface pressure (π) with a Wilhelmy wire attached to a microbalance (DeltaPi, Kibron Inc., Helsinki, Finland) connected to a PC and by using circular glass wells (subphase volume 0.5 mL). After evaporation of solvent and stabilization of monolayers at different initial surface pressures (π0), the peptide (0.1–2 μm) was injected into the subphase, and the increment in surface pressure of the LPS film upon intercalation of the peptide dissolved in the subphase was monitored for the next 37 min. The difference between the initial surface pressure and the value observed after the penetration of Esc(1–18) into the film was taken as Δπ.
Susceptibility testing was performed by the microbroth dilution method according to the procedures outlined by the National Committee for Clinical Laboratory Standards (2001), using sterile 96-well plates. Stock solutions of Esc(1–18) were prepared in serial two-fold dilutions in 20% ethanol; 4 μL was then added to 46 μL of Mueller–Hinton broth, previously placed in the wells of the microtiter plate. Aliquots (50 μL) of bacteria in mid-log phase, at a concentration of 1 × 106 or 4 × 107 CFU·mL−1, were then added to each well.
The range of peptide dilutions used was 1–128 μm. Inhibition of growth was determined by measuring the absorbance at 595 nm with a 450-Bio-Rad Microplate Reader after incubation for 18–20 h at 37 °C. Antibacterial activity was expressed as MIC, the concentration of peptide at which 100% inhibition of growth was observed.
The bactericidal activity of Esc(1–18) against E. coli ATCC 25922 was evaluated by a liquid microdilution assay as described previously , in four different incubation media: sodium phosphate buffer (pH 7.4); Mueller–Hinton broth; NaCl/Pi; and 70% human serum (inactivated by heating at 56 °C for 30 min). Briefly, exponentially growing bacteria were incubated at 37 °C for 1.5 h in the presence of different concentrations of peptide (serial two-fold dilutions ranging from 1 to 128 μm) dissolved in 100 μL of medium. Following incubation, the samples were plated onto LB agar plates. The number of surviving bacteria, expressed as CFUs, was determined after overnight incubation at 37 °C. Bactericidal activity was defined as the peptide concentration necessary to cause a reduction in the number of viable bacteria of ≥ 3 log10 CFU·mL−1 . Controls were run without peptide and in the presence of peptide solvent (20% ethanol) at a final concentration of 0.6%.
About 4 × 106 CFUs in 100 μL of sodium phosphate buffer (pH 7.4) were incubated at 37 °C with Esc(1–18) at the MIC (32 μm) and a subinhibitory concentration (0.25 μm). Aliquots of 10 μL were withdrawn at different intervals, diluted in Mueller–Hinton broth, and spread onto LB agar plates. After overnight incubation at 37 °C, the number of CFUs was counted. Controls were run without peptide and in the presence of peptide solvent (20% ethanol) at a final concentration of 0.6%.
Peptide effect in combination with conventional antibiotics
Combinations of Esc(1–18) and antibiotics with different chemical characteristics, in two-fold serial dilutions in water, were tested for their synergistic effect by a checkerboard titration method. The ranges of drug dilutions used were 0.25–64 μg·mL−1 for Esc(1–18) and 0.25–256 μg·mL−1 for the conventional antibiotics. The mean FIC index for combinations of two peptides was calculated according to the equation
where A and B are the MICs of drug A and drug B in the combination, MICA and MICB are the MICs of drug A and drug B alone, FICA and FICB are the FICs of drug A and drug B and n is the number of wells per plate used to calculate the FIC. The FIC indices were interpreted as follows: FIC ≤ 0.5, synergy; 0.5 < FIC <1, additivity; 1 ≤ FIC < 4, indifference; and FIC ≥ 4, antagonism . The synergistic effect in the bactericidal activity of the combination Esc(1–18) + erythromycin was also determined. E. coli cells (1 × 106 CFUs·mL−1) were incubated in Mueller–Hinton broth (diluted 1 : 2 with distilled water) at 37 °C in the presence of 256 μg·mL−1 erythromycin, a sublethal concentration of erythromycin (8 μg·mL−1) or Esc(1–18) (1 μg·mL−1), and with the combination erythromycin + Esc(1–18) at their sublethal doses. Aliquots were withdrawn at specific time intervals (20, 60, 120 and 180 min), and plated for counting.
Permeation of the bacterial OM
OM permeability was assessed by measuring the activity of the periplasmic β-lactamase. Briefly, E. coli ATCC 25922 cells were grown at 37 °C in Mueller–Hinton broth to a D590 nm of 0.8, and then washed and resuspended in sodium phosphate buffer (pH 7.4) + 100 mm NaCl. About 4 × 106 cells (100 μL of bacterial suspension at a concentration of 4 × 107 CFUs·mL−1) were incubated with different concentrations of peptide (ranging from 4 to 32 μm) for 20 min at 37 °C. The bacterial culture was then passed through a 0.2 μm filter, and a β-lactamase substrate (CENTA, a synthesized chromogenic cephalosporin, with a highly reactive β-lactam ring ) was added to a final concentration of 80 μm. Hydrolysis of the β-lactam ring, which causes a color change from light yellow (λmax: 340 nm) to chrome yellow (λmax: 405 nm), was recorded at 405 nm, using a spectrophotometer (UV-1700 Pharma Spec Shimadzu, Tokyo, Japan). An increase in absorbance results in an increase in OM permeability . The same amount of bacteria without Esc(1–18) was used as a control, whereas the maximal membrane perturbation was obtained after lysing bacteria with 0.1% SDS in chloroform (three drops to 1 mL of bacterial suspension).
Permeation of the bacterial IM
To assess the ability of Esc(1–18) to alter the permeability of the IM of E. coli, 4 × 106 cells were mixed with 1 μm SYTOX Green in NaCl/Pi for 5 min in the dark. After addition of peptide, the increase in fluorescence, owing to the binding of the dye to intracellular DNA, was measured at 37 °C in a microplate counter (Wallac 1420 Victor3™; Perkin Elmer, Foster City, CA, USA), using 485 and 535 nm filters for excitation and emission wavelengths, respectively. The peptide concentrations used ranged from 2 to 64 μm. Controls were cells without peptide. The ability of Esc(1–18) to cause more pronounced damage to the cytoplasmic membrane was determined by measuring the release of β-galactosidase into the culture medium, using Gal-ONp as a substrate . As described above, E. coli cells were grown at 37 °C in Mueller–Hinton broth supplemented with 1 mm isopropyl thio-β-d-galactoside to a D590 nm of ∼ 0.8, and then washed and resuspended in sodium phosphate buffer (pH 7.4). About 4 × 106 cells were incubated with different concentrations of Esc(1–18) for 20 min at 37 °C. Controls were bacteria without peptide, whereas the maximal membrane perturbation was obtained after lysing bacteria with 0.1% SDS in chloroform (three drops to 1 mL of bacterial suspension), as described above. At the end of the incubation time, 2 μL aliquots were withdrawn, diluted 1 : 100 in Mueller–Hinton broth, and spread onto LB plates for counting. The bacterial culture was then passed through a 0.2 μm filter, and the hydrolysis of Gal-ONp was recorded in the culture filtrate at 420 nm using a spectrophotometer (UV-1700 Pharma Spec Shimadzu).
Calcein-loaded and dextran-loaded LUV and leakage assay
Lipid films of PE and PG were prepared by dissolving dry lipids (2 mg of PE/PG mixture, 7 : 3, w/w) in chloroform/methanol (2 : 1, v/v) and evaporating the solvents under a nitrogen stream. The lipid film was then hydrated with 10 mm Tris and 150 mm NaCl (pH 7.4) containing 60 mm calcein solution. The liposome suspension was extruded 10 times through a polycarbonate filter (pore size, 0.1 μm), and free calcein was removed by gel filtration, using a Sephadex G-25 column (1.5 × 10 cm; Pharmacia Biotech AB) at room temperature. Calcein entrapped in the vesicles is highly concentrated, and the fluorescence is self-quenched. Calcein release due to membrane permeation induced by the peptide was monitored at 37 °C by the fluorescence increase (λexcitation = 485 nm; λemission = 535 nm). Complete dye release was obtained using 0.1% Triton X-100, which causes total destruction of lipid vesicles . The apparent percentage of leakage value was calculated according to the following equation : leakage (%) = 100(F1 – F0)/(Ft – F0), where F0 represents the fluorescence of intact vesicles, and F1 and Ft denote the intensities of the fluorescence achieved by peptide and Triton X-100 treatment, respectively.
FITC-D was dissolved in Hepes buffer (10 mm Hepes, 150 mm NaCl, 0.1 mm EDTA, pH 7.4) at a concentration of 4 mm .
The lipid film of PE/PG was resuspended in FITC-D buffer, subjected to several cycles of freezing and thawing, and then extruded as described above. FITC-D LUVs were separated from nonencapsulated dextrans, using a Sephadex G-50 (for FITC-D 4) or Sephadex G-200 (for FITC-D 10, FITC-D 40, and FITC-D 70) gel filtration column . The self-quenching properties of entrapped FITC-D were used in this series of measurements, and peptide-induced dextran leakage was detected after 15 min of peptide treatment at 37 °C, by increases in fluorescence (λexcitation = 470 nm; λemission = 520 nm). Complete leakage was achieved by treating vesicles with 0.1% Triton X-100. The percentage of dye release was evaluated with the equation given above.
CD experiments were performed using a JASCO J-600 spectropolarimeter with a 1 mm path length cell. The CD spectra of the peptide were recorded at 25 °C at 0.2 nm intervals from 195 to 250 nm, at a concentration of 5 μm, in sodium phosphate buffer (pH 7.4) or in a suspension of lipid vesicles composed of PE/PG (7 : 3, w/w), to mimic the E. coli inner membrane , extruded to a diameter of 50 nm. For each spectrum, CD data from eight scans were averaged and expressed as per residue molar ellipticity (θ).
E. coli ATCC 25922 cells were grown in Mueller–Hinton broth to a logarithmic phase, harvested by centrifugation at 1000 g for 10 min, washed twice with 10 mm sodium phosphate buffer (pH 7.4), and then resuspended in the same buffer. About 4 × 106 cells were incubated at 37 °C for up to 20 min with 32 μm Esc(1–18). Controls were run in the presence of the peptide solvent (20% ethanol), at a final concentration of 0.6%. The volume was adjusted to 200 μL, and each sample was spread on a poly(l-lysine)-coated 12 × 12 mm glass slide to immobilize bacterial cells. Glass slides were incubated at 37 °C for 90 min. Slide-immobilized cells were fixed with 2.5% glutaraldehyde in 0.1 m potassium phosphate buffer, extensively washed with the same buffer, and dehydrated with a graded ethanol series. After critical point drying and gold coating, the samples were observed with a Philips XL 30 CP instrument.
Samples containing E. coli ATCC 25922 cells (4 × 106 cells) in sodium phosphate buffer (pH 7.4) were incubated with 32 m Esc(1–18) for 5 and 20 min, and then centrifuged at 300 g for 20 min. Controls were performed in the presence of peptide solvent (20% ethanol), at a final concentration of 0.6%. The pellets were resuspended in sodium phosphate buffer (pH 7.4); a drop containing the bacteria was deposited onto a carbon-coated grid and negatively stained with phosphotungstic acid solution (2%, w/v) (pH 6.8). The grids were examined using a Philips CM 100 electron microscope.
Preparation of E. coli protein extract for 2D-PAGE
E. coli cells (4 × 108 mL−1) were treated with 2 and 16 μm Esc(1–18) in sodium phosphate buffer (pH 7.4) and incubated at 37 °C for 20 min. Treated and untreated cells were harvested by centrifugation at 10 000 g for 15 min at 4 °C, and total protein extract was obtained using the ProteoExtract Complete Bacterial Proteome Extraction kit (Calbiochem cat. 539770), according to the manufacturer’s instructions. The protein concentration was determined by the Bradford assay.
Samples from three independent experiments were analyzed in triplicate. IEF was performed on an Ettan IPG-Phor system (Amersham Biosciences, Uppsala, Sweden), at 16 °C and under a current limit of 50 μA per strip. Sixty micrograms of protein in a final volume of 350 μL of a solution containing 8 m urea, 4% Chaps, 65 mm dithioerythritol (DTE), 0.5% (v/v) ampholine pH 3–10 NL and a trace of bromophenol blue were loaded onto 18 cm pH 3–10 NL Immobiline DryStrip (IPG strip; Amersham Biosciences). The strip rehydration step was performed at 16 °C at a constant voltage of 30 V for 4 h and for an additional 5 h at 50 V. Damp electrode pads were positioned under the rehydrated strip over the electrodes. The IEF step was performed using the following parameters: 400 V for 2 h, 800 V for 1 h, 1200 V for 2 h, 3000 V for 3 h, and 8000 V for 6–8 h, until the total voltage reached 70 kVh. Immediately after the IEF run, IPG strips were equilibrated for 12 min in 6 m urea, 30% (v/v) glycerol, 2% (w/v) SDS, 50 mm Tris/HCl (pH 6.8), and 2% (w/v) DTE, and for 5 min in a similar solution, with a trace of bromophenol blue, in which 2% DTE was replaced with 2.5% (w/v) iodoacetamide. The second dimension of electrophoresis was run on 9–16% linear gradient polyacrylamide gels (18 cm × 20 cm × 1.5 mm) at 40 mA per gel constant current at 10 °C for ∼ 5 h, until the dye front reached the bottom of the gel. Gels were stained with colloidal Coomassie blue; images were acquired on a BioRad GS-800 Calibrated Imaging Densitometer (Bio-Rad, Veenendaal, The Netherlands), and analyzed with the Bio-Rad pdquest software, version 7.1.0. For each spot, an average quantity derived from each replicate group and a coefficient of variation (CV) were calculated. Note that only those spots with a CV ≤ 5% were considered to be ‘valid’ spots for the performance of differential analysis. Spots that were at least two-fold upregulated or downregulated and with a t-test P-value less than 0.05 were considered to be proteins with significantly altered expression, and were thus selected for identification by peptide mass fingerprinting (PMF).
Protein identification by peptide mass fingerprinting (MALDI-TOF MS)
Protein spots were manually excised from the electrophoresis gel, washed with high-purity water and with 50% acetonitrile/water, and then dehydrated with 100% acetonitrile. The gel slices were swollen at room temperature in 20 μL of 40 mm NH4HCO3/10% acetonitrile containing 25 ng·μL−1 trypsin (Trypsin Gold, MS grade; Promega, Madison, WI, USA).
After 1 h, 50 μL of 40 mm NH4HCO3/10% acetonitrile was added, and digestion proceeded overnight at 37 °C. The generated peptides were extracted with 50% acetonitrile/5% trifluoroacetic acid (TFA) (two steps, 20 min at room temperature each), dried by vacuum centrifugation, suspended in 0.1% TFA, passed through microZipTip C18 pipette tips (Milllipore, Bedford, MA, USA), and directly eluted with the MS matrix solution (10 mg·mL−1α-cyano-4-hydroxycinnamic acid in 50% acetonitrile/1% TFA). Mass spectra of the tryptic peptides were obtained using a Voyager-DE MALDI-TOF mass spectrometer (Applied Biosystems). PMF database searching was performed using the mascot search engine (http://www.matrixscience.com) in the ncbinr/swiss-prot databases. Parameters were set to allow one missed cleavage per peptide, a mass tolerance of 0.5 Da, and for carbamido-methylation of cysteines to be considered as a fixed modification and oxidation of methionines as a variable modification. The criteria used to accept identifications included the extent of sequence coverage, the number of matched peptides, and probabilistic score, as detailed in Table 3.
We thank M. Simmaco for use of the facilities and platforms available in the DiMA Unit of the Sant’Andrea Hospital. This work was funded in part by Italian Ministero dell’Università e Ricerca (PRIN 2005 protocol no. 2005062410) and by grants from the Università di Roma La Sapienza and Istituto di Biologia e Patologia Molecolari of the National Research Council.