Design of Potent, Non-Toxic Antimicrobial Agents Based upon the Structure of the Frog Skin Peptide, Temporin-1CEb from Chinese Brown Frog, Rana chensinensis


Corresponding author: Dejing Shang,


Temporin-1CEb shows antimicrobial activity against Gram-positive bacteria, but its therapeutic potential is limited by its haemolysis. In this study, eight temporin-1CEb analogues with altered cationicities and hydrophobicities were synthesized. Increasing cationicity and amphipathicity by substituting neutral and non-polar amino acid residues on the hydrophilic face of the α-helix by five or six lysines increased antimicrobial potency approximately 10-fold to 40-fold, although when the number of positive charges was increased from +6 to +7, the antimicrobial potency was not additionally enhanced. The substitution of an l-lysine with a d-lysine, meanwhile maintaining the net charge and the mean hydrophobicity values, had only a minor effect on its antimicrobial activity, whereas significantly led a decrease in its haemolytic activity. Of all the peptides, l-K6 has the best potential as an antimicrobial agent because its antimicrobial activity against both Gram-positive and Gram-negative bacteria is substantial, and its haemolytic activity is negligible. l-K6 adopts an α-helix in 50% trifluoroethanol/water and 30 mm SDS solutions. l-K6 killed 99.9% of E. coli and S. aureus at 4× MIC in 60 min, and its postantibiotic effect was >5 h. l-K6 affects the integrity of E. coli and S. aureus plasma membranes by rapidly inducing membrane depolarization.

Antimicrobial peptides (AMPs) with broad-spectrum antibacterial activities are synthesized in the skins of frogs and components of the innate immune systems of frogs (1). These compounds have potential as therapeutic agents against microorganisms (2). To date, approximately 200 AMPs, belonging to more than 20 different families, have been identified from more than 40 different amphibians (3–5). Among the AMP families, the members of the temporin family are short (10–13 amino acids), hydrophobic and C-terminally amidated AMPs found in frog skin that have haemolytic activity as well as antimicrobial activity. The first temporin was identified in the skin of the Asian frog Rana erythraea on the basis of its haemolytic activity (6). More than 50 different temporins have subsequently been isolated from the skins of North American and Eurasian frogs of the genus Rana and display potent antimicrobial activity against bacteria, fungi and even certain viruses (7,8). The antimicrobial mechanism of temporins is thought to be based on the leakage of water and ions through pores in the bacterial cell membrane (9). Although the use of the naturally occurring temporin-1CEb as antibiotics is limited because of their substantial haemolytic activities against human erythrocytes, temporins modified according to amphipathicity, charge and secondary structure may still represent promising drug candidates.

Temporin-1CEb is a naturally occurring 12-residue peptide derived from skin secretions of Rana amurensis (the peptide is also known as amurin-3) and Rana dybowskii (also known as temporin-CDYb) (10,11). Temporin-1CEb exhibits strong hydrophilicity and no positive charges. Previously, we cloned the cDNA of the temporin-1CEb precursor (EU746505) from the Rana chensinensis, and the synthetically replicated peptide was shown to be moderately high potency against the Gram-positive (MICs: 62 μm) but was only very weakly active against Gram-negative bacteria (MICs: 125 μm) and to have substantial haemolytic activity (LD50: 112 μm) (12). antheprot 4.3 (Institute of Biology and Chemistry of Proteins, France) predicted that temporin-1CEb can assume α-helices of weak amphipathic character (Figure 2). The aim of this study is to design temporin-1CEb analogues that maintain α-helical character and amino acid sequence, but having increased cationicity and decreased hydrophobicity among the helical residues, decreased haemolytic activity and increased activity against various pathogenic microorganisms. To do so, neutral and non-polar amino acid residues on the hydrophilic face of the helix were replaced with l-lysine. In addition to the lysine substitutions, the hydrophobicities of the eight analogues were modified to reflect changed haemolysis. We measured the antimicrobial activities of temporin-1CEb and its analogues against Gram-positive and Gram-negative bacteria; haemolytic activities for all peptides were measured using human erythrocytes. Additionally, for l-K6 (the analogue that displayed the best antimicrobial activity and least haemolysis), its antibactericidal activity and effects on Escherichia coli and Staphylococcus aureus membranes were assessed by determining its kinetic kill curve and its ability to disrupt membrane integrity and cause membrane depolarization.

Material and Methods

Purification of temporin-1CEb

Rana chensinensis adults were captured in the north-eastern region of Liaoning Province, People’s Republic of China. Skin secretions were obtained by mild transdermal electrical stimulation (12). The secretions were frozen in liquid nitrogen, lyophilized, and stored at −80 °C. The frogs were not harmed by this procedure and were later returned to their natural habitat in good health. The lyophilized secretions were dissolved in 0.1% (v/v) trifluoroacetic acid (TFA)/water and applied to a reversed-phase HPLC Vydac 218 TP1010 (C-18) column (1 × 25 cm; Separations Group, Hesperia, CA, USA) that had been equilibrated with 0.1% (v/v) TFA/water. The linear gradient eluent was from 0 to 50% (v/v) acetonitrile/water (constant 0.1% v/v TFA) and developed over 90 min. The flow rate was 3.3 mL/min. Fractions with antimicrobial activity were collected and lyophilized, respectively. Then, each of lyophilized fractions was dissolved in 0.1% (v/v) TFA/water and chromatographed through a Vydac 214 TP 510 (C-4) column (1 × 25 cm) with a 0–60% gradient of acetonitrile/water (constant 0.1% v/v TFA) at a flow rate of 0.5 mL/min for 60 min. The purified fraction was collected and lyophilized.

Structural characterization

Automated Edman degradation (Procise Sequencer, Model 494; Applied Biosystems, Foster City, CA, USA) and mass spectrometry (MS) (Q-TOF2 mass analyser; Micromass Ltd., Manchester, UK) were used to sequence temporin-1CEb. The relative mass of temporin-1CEb was obtained from a matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrum recorded with a Voyager instrument (Applied Biosystems). The physicochemical features of temporin-1CEb and its analogues were obtained from

Synthesis of temporin-1CEb analogues

The temporin-1CEb analogues were synthesized in crude form by the standard Fmoc solid-phase peptide synthesis protocols of GL Biochem Ltd. (Shanghai, China) and were purified to near homogeneity (>95%) by reverse-phase HPLC. The relative masses of the analogues were obtained using HPLC-(enhanced product ion scan) MS and MALDI-TOF MS (Shimadzu, Japan).

Effective hydrophobicity and amphipathicity of the peptides

The effective hydrophobicity of the peptide was monitored by RP-HPLC (13,14). Peptides were run on a reversed-phase HPLC Vydac 218 TP54 (C-18) column (0.46 × 25 cm; Separations Group, Hesperia, CA, USA) that had been equilibrated with acetonitrile/water/TFA (18.0/81.9/0.1). The concentration of acetonitrile was raised to 56.0% over 50 min using a linear gradient. The flow rate was 2.0 mL/min. Absorbance was measured at 214 nm, and the retention time of each peptide was recorded. The mean hydrophobicity values of peptide used in this study were calculated using hydrophobicity scales (15) and were listed as follows: Trp, 32.4; Phe, 29.1; Leu, 23.3; Ile, 21.4; Met, 15.7; Tyr, 14.7; Val, 13.4; Pro, 9.0; Cys, 7.6; Lys, 2.8; Glu,2.8; Ala, 2.8; Thr, 2.3; Asp, 1.6; Arg, 0.6; Gln, 0.6; His, 0; Ser, 0; Gly, 0; Asn, -0.6. Amphipathicity values of all analogues were calculated by hydrophobic moment (16), using the software package Jemboss version 1.5 (17).

Antimicrobial assay

The bacterial strains E. coli (AS 1.349), Pseudomonas aeruginosa (CGMCC 1.860), Enterobacter aerogenes (CGMCC 1.876), Enterobacter cloacae (CGMCC 1.58), Klebsiella pneumoniae subsp. Pneumoniae (CGMCC 1.176), Staphylococcus aureus (AS 1.72), Bacillus cereus (AS 1.126), Streptococcus lactis (AS 1.1690), Enterococcus faecalis (CGMCC 1.595) and Enterococcus faecium (CGMCC 1.2334) were acquired from the China General Microbiological Culture Collection Centre. The antimicrobial activity of each peptide was tested using an inhibition zone assay on agarose plates seeded with viable bacteria (18). MICs for the bacteria were measured using a standard micro-dilution method with 96-well microtitre plates (19). Briefly, each peptide was serially diluted to concentrations between 1.0 and 1000 μm in Mueller–Hinton broth. Then, 50-μL samples were dispensed into the wells of the plates, and each sample was mixed with a 50-μL inoculum (10CFU/mL) of a log-phase bacterial culture. The absorbance of each sample at 600 nm was recorded using a microtitre plate reader after the cultures had been held for 18–24 h at 37 °C. The MIC was defined as the smallest peptide concentration that completely inhibited bacterial growth. To assess the precision of the assay, parallel incubations in the presence of gentamycin sulphate were performed.

Haemolysis assay

The haemolytic activities of the peptides were determined by the method of Lequin et al. (20). Briefly, 2 × 107 erythrocytes from the blood of healthy humans were washed three times with 0.9% NaCl and then incubated with a peptide (final concentration, 1–1000 μm) for 3 h at 37 °C. After centrifugation, the absorbance (545 nm) of each resuspended pellet solution was measured. The positive control was an erythrocyte suspension incubated in water (100% haemolysis), while the negative control was an erythrocyte suspension incubated in 0.9 NaCl (0% haemolysis). The LD50 was defined as the mean peptide concentration from three independent experiments that produced 50% haemolysis.

Circular dichroism spectroscopy

Circular dichroism (CD) spectra of temporin-1CEb and l-K6, between 190 and 250 nm, were recorded at 0.1-nm intervals, at 25 °C, and at a scan rate of 20 nm/min using a J-810 spectropolarimeter (JASCO, Victoria, British Columbia, Canada). Peptides (0.3 mg/mL) were dissolved in water, or 30 mm SDS solution, or 50% (v/v) trifluoroethanol (TFE) in water. A quartz cuvette with a 1-mm path length was used. Three consecutive scans were averaged, and the CD spectrum of each solvent (minus the peptide) was subtracted from the corresponding peptide spectrum.

Bactericidal kinetics of l-K6

The bactericidal activity of l-K6 was assessed using time-kill curves. Briefly, l-K6 was added at final concentrations of 0.5×, 2× and 4× its MIC value to log-phase cultures (106 CFU/mL) of E. coli and S. aureus that were then incubated with shaking at 37 °C in a water bath for 0–180 min. Each sample was then washed twice with sterile Mueller–Hinton broth. The surviving bacteria were measured using a spiral-plating system (Don Whitley Scientific, Shipley, UK) after being diluted 102- or 104-fold. A control (no l-K6) was run under the same conditions. Bactericidal kinetics were determined by plotting the number of surviving bacteria against time. The absolute limit of detection was 2 × 101 CFU/mL.

Postantibiotic effect of l-K6

The postantibiotic effect (PAE) measures the capacity of an antimicrobial drug, once taken up by the microbe inhibit the growth of bacteria after it has been removed from the culture. Briefly, l-K6 samples (final concentration, 0.5×, 2× or 4× its MIC value) were individually mixed with log-phase cultures (106 CFU/mL) of E. coli or S. aureus. After a 60-min incubation, each culture was washed twice with Mueller–Hinton broth. The cultures resuspended in a liquid medium were incubated under shaking in a water bath at 37 °C for 8 h. During that period of time, the cell numbers were determined at hourly intervals using the spiral-plating method. The same procedure was applied to the control (no l-K6) experiment. The time that took for the colonies to reach a CFU of 1.0log10 was recorded. PAE was defined as the time difference between an experimental culture and the control culture with an increase of 1.0log10 CFU/mL.

Assessing bacterial membrane integrity in the presence of l-K6

Bacterial membrane integrity was determined with reagents from the LIVE/DEAD BacLight Bacterial Viability kit L7012 (Molecular Probes Inc., Eugene). Briefly, log-phase cultures of E. coli or S. aureus were centrifuged, and the pellets were resuspended in 0.85% NaCl. The suspensions were mixed with l-K6 (final concentration, 0.5×, 2× or 4× its MIC value) and incubated for 1 h. After washing, the cells were resuspended in 0.85% NaCl to yield 2 × 107 bacteria/mL. Next, the dye solution (an appropriate mixture of the SYTO 9 and propidium iodide stains) was added, and the mixtures were incubated in the dark for 15 min. The fluorescence of the dye was measured spectrofluorometrically: excitation at 485 nm + emission at 530 nm (emission 1; green), and excitation at 485 nm + emission at 630 nm (emission 2; red). The reported RatioG/R values are a measure of the membrane integrity and were obtained by dividing the green fluorescence intensity of emission 1 by the red intensity of emission 2. The positive control was a bacterial suspension incubated in 70% isopropyl alcohol (which killed the bacteria), and the negative control was a bacterial suspension in 0.85 NaCl.

Assessment of bacterial membrane depolarization by l-K6

The extent of plasma membrane depolarization was measured using the membrane potential-sensitive fluorescent dye diSC3-5 (Sigma-Aldrich). Briefly, E. coli and S. aureus were grown to mid-log phase, centrifuged, washed twice with 5 mm HEPES, pH 7.2, containing 20 mm glucose and 100 mm KCl and resuspended to an OD600 of 0.05 in the same solution. For the solution containing E. coli, EDTA was also added (final concentration, 0.5 mm). Then, diSC3-5 was added (final concentration, 4 μm). The fluorescence of the dye was monitored at 1-min intervals (excitation 622 nm; emission 670 nm). After the maximum amount of dye had been taken up the bacteria, l-K6 was added at a final concentration of 0.5×, 2× or 4× its MIC value to samples of the bacteria, and membrane permeability was determined by the variation in fluorescence intensity, which was recorded at 1-min intervals. Measurements were repeated at least three times for each dye concentration.


Isolation and structural characterization of temporin-1CEb

Rana chensinensis skin secretions were chromatographed through a Vydac C18 reversed-phase HPLC column (Figure 1). Each fraction was screened for antimicrobial activity by the inhibition zone assay. To shorten the retention time, the fraction that displayed antimicrobial activity was purified to near homogeneity by chromatography through a Vydac C4 reversed-phase HPLC column. The sequence of temporin-1CEb was determined to be ILPILSLIGGLL-NH2 by automated Edman degradation and quadrupole-TOF MS. The amino acid sequence was the same as that previously deduced from the mature peptide of preprotemporin-1CEb (12). The C-terminal amino acid sequence of preprotemporin-1CEb deduced from the cDNA sequence was Gly-Lys. Hydrolysis by carboxypeptidase during processing of these proforms would expose the glycine residue, which is required for the formation of C-terminal amides. The calculated mass of one peptide with C-terminally amidation should be 1 Da less than the peptide without amidation (21). The observed mass of temporin-1CEb (1219.94), determined by MALDI-TOF MS, was 1 Da less than predicted from the amino acid sequence deduced by Edman degradation thus implying C-terminal amidation.

Figure 1.

 Preparative reverse-phase HPLC (Vydac C-18 column) of skin secretions from R. chensinensis. The elution position of temporin-1CEb is indicated by an arrow. The dashed line shows the acetonitrile/water gradient.

Design of the temporin-1CEb analogues

To enhance the cationicity of temporin-1CEb, l-K5 and l-K6 were synthesized with l-Lys substituted for Leu2, Pro3, Leu7, Gly9 and Gly10; in the case of l-K6, an additional l-Lys was inserted between the C-terminal Leu and amide. These substitutions increased the net charge of l-K5 and l-K6 from +1 (temporin-1CEb) to +6 and +7, respectively. antheprot 4.3 (Institute of Biology and Chemistry of Proteins, France) predicted that l-K5 and l-K6 can assume α-helices of considerable amphipathic character with the Lys and Ser6 segregated on one face and the hydrophobic residues in l-K5 and l-K6, Ile1, Ile4, Leu5, Ile8, Leu11 and Leu12 segregated on the opposite face (Figure 2). To examine the effects of Val on antimicrobial activity and haemolysis, three additional peptides (l-K5V1, l-K6V1 and l-K6V2) were prepared by replacing Leu5 with a Val in l-K5 and l-K6, or replacing Leu5 and Leu11 with Val in l-K6, respectively. For d-K5V2, Val replaced Leu11 and d-Lys replaced l-Lys7 in l-K5V1. d-K6V2 was designed by the replacing the l-Lys2 of l-K6V2 with d-Lys. Table 1 lists the sequences of the peptides, their observed and calculated molecular masses, net charges, retention time determined by RP-HPLC, mean hydrophobicity values (H) of the peptides based upon the hydrophobicity scales for amino acids and amphipathicity values calculated by hydrophobic moment using jemboss software. All analogues are strongly cationic, with a net positive charge of +6 or +7. Temporin-1CEb is dramatically more hydrophobic than all the analogues that have 5 or 6 Lys residues to replace Leu2, Pro3, Leu7, Gly9 and Gly10 in the case of l-K5, l-K5V1, d-K5 and d-K5V2, or an additional Lys residue in l-K6, l-K6V1, l-K6V2 and d-K6V2. The hydrophobicity changes from the retention time of 39.1 min to a range of 4.8–12.6 min for the eight analogues. Interestingly, the substitution of two Leu residues in d-K5 with two Val residues in d-K5V2 results in the observed hydrophobicity actually increases from the retention time of 10.7–12.6 min, which is the opposite to that predicted from the calculated hydrophobicities where the H values are 12.3 for d-K5 (which as two Leu residues) and 10.7 for d-K5V2 (two Val residues). The results suggest that the β-branched Val residues have changed the conformation of the peptide relative to Leu residues increasing the observed hydrophobicity. The amphipathicity values of the eight analogues increase significantly from 0.382 to 0.829 (or 0.830) compared with temporin-1CEb. However, a slight change among these analogues was observed.

Figure 2.

 Helical diagrams for temporin-1CEb, l-K5 and l-K6. 1, temporin-1CEb; 2, l-K5; 3, l-K6; The diagrams were generated using antheprot 4.3 (Institute of Biology and Chemistry of Proteins, France).

Table 1.   Amino acid sequences and physical characteristics for temporin-1CEb and its analogues
Antimicrobial peptidesSequenceaMass calc./obs.b (Da)Net chargeRTc(min)HdAmphipathicitye
  1. aK indicates d-Lys.

  2. bMass obs. /calc represented observed molecular masses determined by MALDI-TOF MS and calculated molecular masses based on the amino acid sequence determined by Edman degradation.

  3. cThe retention time was measured on a Vydac 218TP54 column using the elution conditions described in Materials and Methods.

  4. dThe mean hydrophobicities (H) of the peptides calculated using the hydrophobicity scales (15) were the total hydrophobicity (sum of all residue hydrophobicity indices) divided by the number of residues.

  5. eAmphipathicity was determined by calculation of hydrophobic moment (16,17).


Antimicrobial and haemolytic activities of temporin-1CEb and its analogues

The MICs of temporin-1CEb and its analogues for tested bacteria and their LD50 values for the lysis of human erythrocytes are listed in Table 2. All analogues had substantially improved antimicrobial activity against all of the tested organisms when compared with those of temporin-1CEb, and their haemolytic activities were also much improved. l-K5, l-K6, l-K5V1 and l-K6V1 were the most potent analogues (MIC ≤ 6.25 μm for both Gram-positive and Gram-negative bacteria—a minimum of a 10–30-fold increase compared with temporin-1CEb), and their haemolytic activities varied but were all lower than that of temporin-1CEb. l-K6 and l-K6V1 (the observed hydrophobicity is the retention time of 4.8 and 5.0 min, respectively) did not lyse the erythrocytes (LD50 values > 1000 μm), whereas l-K5 and l-K5V1 (the observed hydrophobicity is the retention time of 11.0 and 11.5 min, respectively) did (LD50 values of 550 and 352 μm, respectively). d-K5 and d-K6V2 retained high potency against the Gram-positive bacteria (MIC 3.12–6.25 μm) but slightly decreased activity against Gram-negative bacteria strain (MIC 6.25–7.81 μm) as compared with l-K5 and l-K6V2, in which the l-Lys7 and l-Lys2 were replaced by d-Lys7 and d-Lys2, respectively. However, the haemolytic activities of d-K5 and d-K6V2 significantly decreased (LD50 values increased from 550 μm (and 384 μm) to >1000 μm). The single Val-substituted analogue l-K6V1 had no significant effect on the activity against Gram-positive and Gram–negative bacteria and the haemolysis. However, the introduction of two Val residues at positions 5 and 11 (l-K6V2) led to decreases in potency against Gram-positive and Gram–negative bacteria, meanwhile, increase in the haemolytic activity from LD50 > 1000 to 384 μm, as compared with l-K6.

Table 2.   MIC values (μm) and haemolytic activities (LD50, μm) of temporin-1CEb and its analogues
 MIC (μm)*
  1. *MIC, minimal peptide concentration required for total inhibition of cell growth in liquid medium. Values represent the means of three independent experiments performed in duplicate.

G bacterium
 E. coli1253.
 P. aeruginosa1253.
 E. aerogenes2506.
 E. cloacae2506.
 K. pneumoniae1756.
G+ bacterium
 S. aureus62.
 B. cereus62.
 S. lactis62.
 E. faecalis1256.
 E. faecium1256.

CD spectra of temporin-1CEb and l-K6

The secondary structures of temporin-1CEb and l-K6 in water, or 30 mm SDS, or 50% TFE/water, which mimicked the cell membrane environment, were assessed by CD spectroscopy. The conformations of the peptides changed substantially from a random coil–like structure in water to α-helical-like structures in 30 mm SDS and 50%TFE/water (Figure 3). The spectra for temporin-1CEb and l-K6 in the helix-inducing solvents had negative minima at 208 and 220 nm, which are typical for α-helices. Using the molar ellipticity at 222 nm, the calculated helical content for L-K6 is 75% and 87.5% in TFE or in SDS solution, respectively. Given their spectra, both l-K6 and temporin-1CEb were most helical in SDS solution.

Figure 3.

 CD spectra of temporin-1CEb and l-K6 in water, 50% trifluoroacetic acid/water and 30 mm SDS. All peptides are unstructured in water alone, whereas adopt α-helical structure in the presence of trifluoroethanol or of SDS micelles, mimetic of a membrane environment.

Killing kinetics of l-K6

l-K6, which had substantial antimicrobial activity and no haemolytic activity, was used as the model to characterize bactericidal activity. Time killing kinetics of l-K6 showed a clear dose–response relationship. l-K6 rapidly killed the bacteria, with a > 4log10 reduction in the number of viable cells after only 60 min at concentrations of 2× and 4× its MIC value (Figure 4). There was no significant change in the CFUs for all three l-K6 concentrations between 60 and 180 min. Owing to the rapid bactericidal activity of l-K6, a long PAE period was observed for both the E. coli and S. aureus samples (6 h at both 2× and 4× its MIC value) after a 60-min exposure to l-K6 (Figure 5).

Figure 4.

 Killing kinetics of l-K6 at 0.5×, 2× and 4× its MIC value for (A) E. coli and (B) S. aureus.l-K6 was added to log-phase cultures of E. coli and S. aureus and then incubated for 0–180 min. The surviving bacteria were measured using a spiral-plating system after being diluted 102- or 104-fold. A control (no l-K6) was run under the same conditions. Each data point is the average of six independent experiments ± the standard error of the mean.

Figure 5.

 Postantibiotic effect of l-K6 at 0.5×, 2× and 4× its MIC value for (A) E. coli and (B) S. aureus. The log-phase cultures of E. coli or S. aureus was individually mixed with l-K6. After a 60-min incubation, each culture was washed twice and then incubated for 8 h during which time the cell numbers were determined at hourly intervals using the spiral-plating method. The same procedure was applied to the control (no l-K6). Each data point is the average of six independent experiments ± the standard error of the mean.

Effect of l-K6 on the plasma membranes permeabilization

We assessed the ability of l-K6 to damage the cytoplasmic membranes of E. coli and S. aureus by monitoring the influx of the green-fluorescent nucleic acid stain, SYTO 9, and the red-fluorescent nucleic acid stain, propidium iodide. The results demonstrated that l-K6 enhanced the bacterial membrane permeability in a concentration-dependent manner (Figure 6). Treatment with 1.5 μm l-K6 decreased 30% of RatioG/R values comparable with the treatment with 0.85% NaCl. However, even at 4× its MIC value (12 μm), l-K6 did not completely destroy the structure of the membrane as the RatioG/R values were smaller in the case of the bacteria lysed in 70% isopropyl alcohol.

Figure 6.

 Effect of l-K6 on the integrity of the E. coli and S. aureus plasma membrane. The l-K6 concentrations were 0.5×, 2× and 4× its MIC value. The positive control was 70% isopropyl alcohol (to kill all bacteria), and the negative control was 0.85 NaCl. Each value represents the mean of three independent experiments. *p ≤ 0.05 compared with the negative control group. **p ≤ 0.01 compared with the negative control group.

Effect of l-K6 on membrane potential

The membrane potential-sensitive dye diSC3-5 is used to monitor the plasma membrane depolarization of E. coli and S. aureus in the presence of peptide. As shown in Figure 7, after diSC3-5 was added to bacteria cells, the fluorescence decreased as the dye accumulated in the membranes as a result of self-quenching. A significantly increased fluorescent signal was observed right after the addition of l-K6, and the levels of diSC3-5 release were concentration-dependent. l-K6 exhibited higher depolarization capacity in S. aureus than gramicidin D. Importantly, l-K6 is able to cause rapid depolarization of the cytoplasmic membrane of E. coli and S. aureus within 1 min at a concentration of 0.5 × MIC. This concentration will not result in a significant decrease in bacterial viability.

Figure 7.

 Plasma membrane depolarization of (A) E. coli and (B) S. aureus. The membrane potential–sensitive dye diSC3-5 was added to E. coli or S. aureus cells at 4 μm, and the change in dye fluorescence (excitation at 622 nm, emission at 670 nm) was monitored until it was stable. l-K6 was added to samples of the bacteria at 10 or 6 min, and the disruption of the membrane potential as measured by the increase in fluorescence was assessed. The increased fluorescence intensity associated the addition of gramicidin D, which caused the complete disruption of the plasma membrane potential, served as the positive control.


Antimicrobial peptides have great potential to form a new class of antibiotics, because of their broad-spectrum activity, rapid action and low levels of induced resistance. According to the widely accepted two-step mechanism of AMPs, the initial step is the binding of AMPs to the bacterial plasma membrane surface. Once bound, the helical amphipathic structure of peptide can then promote their insertion into the membrane, thereby causing transient pore formation and membrane disruption (22,23). The bacterial plasma membrane is rich in acidic phospholipids, and the cell wall contains negatively charged lipopolysaccharides in gram-negative bacteria and negatively charged teichoic and teichuronic acids in gram-positive bacteria (24,25). Hence, an increase in peptide cationicity can promote the interaction with more negatively charged bacterial cell membrane and thus increase antimicrobial potency. The hypothesis is strongly supported by that we can increase the antimicrobial activity of temporin-1CEb by 10-fold to 40-fold by increasing its net positive charge from +1 to +6. Whereas, in agreeing with the previous studies, that is, using analogues of magainin-2 and pseudin-2 (26,27), further increases in the cationicity (net positive charge from +6 to +7) do not result in any additional increases in antimicrobial activity.

The substitutions of neutral and non-polar amino acid residues on the hydrophilic face of the helix with l-Lys can not only successfully increase cationicity and amphipathicity of AMPs but also decrease hydrophobicity concomitant with increased α-helicity (28,29). The studies of these analogues of magainin 2, ascaphin-8 and peptide XT-7 have demonstrated a direct correlation between increasing hydrophobicity and haemolysis (25,30). Our results also showed that the substitution of selected hydrophobic residues by lysine might produce significant changes in an effective hydrophobicity of the peptides as indicated by their chromatographic properties and the decreases in haemolytic activity accompanied by maintaining or increasing the antimicrobial potency. When the observed hydrophobicity by RP-HPLC decreased from the retention time from 39 to 11 min, the haemolysis of the analogues of temporin-1CEb decreased 2–3 times (LD50 values from 150 to 352 μm or 550 μm) comparable with temporin-1CEb, and the haemolysis decreased more than 10-fold (LD50 values > 1000 μm) when the observed hydrophobicity decreased continuously to the retention time of from 4 to 6 min. Previous studies with model amphipathic α-helical peptides have shown that selective substitutions by d-amino acids may produce analogues that retain antimicrobial activity but show reduced haemolytic activity (31,32). Similar to theses studies, in this study, the inclusion of d-Lys in d-K5 and d-K6V2 had only minor effects on their antimicrobial activities but significantly decreased their haemolytic activities. As there were no differences in the net charge and the calculated hydrophobicities between d-K5 (d-Lys7) and l-K5 (l-Lys7), d-K6V2 (d-Lys2) and l-K6V2 (l-Lys2), respectively, but the observed hydrophobicities are different. This suggests that the switch in stereochemistry does make a subtle difference in overall hydrophobicity, resulting in the disparity of haemolysis activity between the two peptides. Lee and co-workers have synthesized an artificial liposome composed of PC:cholesterol (10:1, w/w) and treated with the peptides l-pleurocidin or d-pleurocidin. They demonstrated that d-pleurocidin did not make a pore on the membrane potently and could not damage human erythrocytes (33). Compared with l-K5 and l-K6, the only change is the addition of an extra Lys at the C-terminal in l-K6, which dramatically decreases the observed overall hydrophobicity from the retention time of 11.0–4.8 min and the H values varied from 12.3 to 11.6, respectively. The large decrease in the hydrophobicity whereas together with the small decrease in the calculated values suggest that the additional Lys residue affects the conformation of the peptide much more that suggested by the calculated hydrophobicity.

l-K6 has the best potential to become an antimicrobial agent because it effectively killed both Gram-positive and Gram-negative bacteria and had no observable haemolytic activity. In this study, the interaction of l-K6 with the bacterial membranes of S. aureus and E. coli was investigated to understand the differing mechanisms by which a single peptide may interact with Gram-positive and Gram-negative bacteria. In common with other amphipathic cationic peptides, l-K6 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, l-K6, adopted an α-helical conformation. Time-kill kinetics against E. coli and S. aureus demonstrated that l-K6 rapidly killed the bacteria at a low concentration and that its PAE was >5 h. A long PAE is beneficial as it provides additional time for the immune system to remove bacteria that might have survived after the drug had been removed (34), which can therefore influence the clinical outcome of antimicrobial therapy. The ability of AMPs to depolarize the bacteria plasma membrane provides a direct assessment of their effect at the membrane inner level (35,36). l-K6 also affected the integrity of the plasma membrane of E. coli and S. aureus by inducing membrane depolarization in <1 min. However, there were temporal differences between membrane depolarization and loss of viability, with depolarization occurring more rapidly. Therefore, it appears that the loss of cell membrane integrity is the cause of the bactericidal activity of l-K6. However, other intracellular processes cannot be ruled out, including disruption of macromolecular synthesis and/or reversible protein phosphorylation, or energy-transducing processes (25,37,38).

Conclusions and Future Directions

Antimicrobial peptides represent a potential source of new antibiotics. To design new AMPs analogues with modifications to the cationicity and hydrophobicity is helpful to understand how these modifications affect the antimicrobial activity and selectivity of AMPs. l-K6, an analogue of temporin-1CEb, may find use as a new drug candidate because it displayed the most improved antimicrobial activity against Gram-positive and Gram-negative bacteria and showed greatly decreased haemolytic activity in comparison with temporin-1CEb.


This work was supported by the National Natural Science Foundation of China (Grant No. 30970352), the Liaoning Key Laboratory Program (2008S131), the Liaoning Excellent Talents in University Program (2007R27) and a Science Grant (2011E12SF031) from the Dalian Science and Technology Bureau.

Competing Interests

None declared.


  •  Eight analogues of temporin-1CEb are designed to understand the relationship of structure-activity by modifications of the cationicity and hydrophobicity.
  • l-K6 may find use as a new drug candidate because it displayed the most improved antimicrobial activity against Gram-positive and -negative bacteria and showed greatly decreased haemolytic activity.
  • l-K6 affected the integrity of the plasma membrane of E. coli and S. aureus by inducing membrane depolarization in <1 min.