Single Amino Acid Substitutions at Specific Positions of the Heptad Repeat Sequence of Piscidin-1 Yielded Novel Analogs That Show Low Cytotoxicity and In Vitro and In Vivo Antiendotoxin Activity

Abstract

Piscidin-1 possesses significant antimicrobial and cytotoxic activities. To recognize the primary amino acid sequence(s) in piscidin-1 that could be important for its biological activity, a long heptad repeat sequence located in the region from amino acids 2 to 19 was identified. To comprehend the possible role of this motif, six analogs of piscidin-1 were designed by selectively replacing a single isoleucine residue at a d (5th) position or at an a (9th or 16th) position with either an alanine or a valine residue. Two more analogs, namely, I5F,F6A-piscidin-1 and V12I-piscidin-1, were designed for investigating the effect of interchanging an alanine residue at a d position with an adjacent phenylalanine residue and replacing a valine residue with an isoleucine residue at another d position of the heptad repeat of piscidin-1, respectively. Single alanine-substituted analogs exhibited significantly reduced cytotoxicity against mammalian cells compared with that of piscidin-1 but appreciably retained the antibacterial and antiendotoxin activities of piscidin-1. All the single valine-substituted piscidin-1 analogs and I5F,F6A-piscidin-1 showed cytotoxicity greater than that of the corresponding alanine-substituted analogs, antibacterial activity marginally greater than or similar to that of the corresponding alanine-substituted analogs, and also antiendotoxin activity superior to that of the corresponding alanine-substituted analogs. Interestingly, among these peptides, V12I-piscidin-1 showed the highest cytotoxicity and antibacterial and antiendotoxin activities. Lipopolysaccharide (12 mg/kg of body weight)-treated mice, further treated with I16A-piscidin-1, the piscidin-1 analog with the highest therapeutic index, at a single dose of 1 or 2 mg/kg of body weight, showed 80 and 100% survival, respectively. Structural and functional characterization of these peptides revealed the basis of their biological activity and demonstrated that nontoxic piscidin-1 analogs with significant antimicrobial and antiendotoxin activities can be designed by incorporating single alanine substitutions in the piscidin-1 heptad repeat.

INTRODUCTION

Fish antimicrobial peptide (AMP) piscidin-1, which was discovered in 2001, possesses versatile biological activities. Piscidin-1 shows significant activity against bacteria, fungi, parasites, and cancer cells (1–7). It can also neutralize lipopolysaccharide (LPS)-induced proinflammatory responses in macrophage cells (5). Along with these desired biological activities, piscidin-1 also exhibits very significant lytic activity against normal mammalian cells, which is an obstacle for employing it as a lead molecule for the development of a new antimicrobial agent. Therefore, deciphering of the basis of cytotoxicity in piscidin-1 and the design of nontoxic analogs of piscidin-1 with desired biological activity were the objectives of the present investigation. Toward this end, we intended to identify the important primary sequence in piscidin-1 that could have a strong impact on its structural, functional, and biological properties. After carefully looking into the sequence of piscidin-1, we identified in it a long heptad repeat sequence which is located in the region from amino acids 2 to 19 and which has not been reported before, to our knowledge. The heptad repeat sequence consists of 7 amino acids (amino acids a to g) where each a and d position is occupied by a hydrophobic residue, such as leucine, isoleucine, or phenylalanine. The role of this sequence in maintaining cytotoxicity has also been examined in other antimicrobial peptides (8–10). The leucine, isoleucine, and phenylalanine residues positioned at the a and d positions of a heptad repeat sequence could interact with similar amino acids of another heptad repeat and thus assist in the self-assembly of a peptide. Reports from several research groups suggest that self-assembly of an antimicrobial peptide greatly influences its cell selectivity/cytotoxicity (8, 11–14). Substitutions of hydrophobic amino acids at the a and/or d position of a heptad repeat sequence can alter the assembly of a peptide containing this motif and also significantly diminish its cytotoxicity (10).

However, how the substitution of amino acids at the a and d positions of a heptad repeat sequence alters the cytotoxic, antimicrobial, and antiendotoxin properties of piscidin-1 has not been addressed before. Further, it is challenging to obtain an analog of piscidin-1 with the desired antiendotoxin properties but reduced cytotoxicity, considering the overlap in the structural requirements of these two biological properties of an antimicrobial peptide (15). If it is assumed that the hydrophobic amino acids located at these a and d positions can also play a prominent role in the peptide-LPS interaction, only single amino acid substitutions would need to be made at these specific positions over the entire length of the identified heptad repeat of piscidin-1. The isoleucine residues at two a positions and one d position of the heptad repeat were individually replaced by three alanine and three valine residues. In one of the single alanine-substituted analogs, namely, I5A-piscidin-1, the alanine residue at the d (5th) position was replaced with a phenylalanine residue located in the adjacent (6th) position without changing its original amino acid composition, resulting in the new analog I5F,F6A-piscidin-1. Another analog (V12I-piscidin-1) was designed by replacing the valine residue at the 12th position (also a d position) of piscidin-1 with an isoleucine residue. Thus, altogether eight analogs of piscidin-1 were designed. Piscidin-1 and its analogs were synthesized, and their antimicrobial activities against different microorganisms, cytotoxicities against mammalian cells, and antiendotoxin properties in the human monocytic cell line THP-1 were characterized. To understand the basis of the biological activities of these peptides derived from piscidin-1, related structural and functional studies were also carried out. The results presented here illustrate a crucial role of the heptad repeat sequence that was identified in controlling the cytotoxic and antiendotoxin activities of piscidin-1. Further, the present study describes the utilization of this heptad repeat sequence for the design of nontoxic analogs of piscidin-1 with significant activity against microorganisms and the ability to neutralize LPS-induced proinflammatory responses in THP-1 cells in vitro and in vivo in mice.

MATERIALS AND METHODS

Rink amide 4-methylbenzhydrylamine (MBHA) resin (loading capacity, 0.4 to 0.8 mmol/g), N-α-9-fluorenylmethoxy carbonyl (Fmoc), and necessary side chain-protected amino acids were procured from Novabiochem. The coupling reagents for peptide synthesis, which included 1-hydroxybenzotriazole (HOBT), di-isopropylcarbodiimide, 1,1,3,3-tetramethyluronium tetrafluoroborate (TBTU), and N,N′-diisopropylethylamine (DIPEA), were obtained from Sigma, USA. Dichloromethane, N,N′-dimethylformamide (DMF), and piperidine were of standard grades and purchased from reputable local companies. Acetonitrile of high-performance liquid chromatography (HPLC) grade was procured from Merck, India. Trifluoroacetic acid (TFA), HEPES, sodium dodecyl sulfate (SDS), Alexa Fluor-annexin V, valinomycin, dimethyl sulfoxide (DMSO), and cholesterol (Chol) were purchased from Sigma. Egg phosphatidylcholine (PC) and egg phosphatidylglycerol (PG) were obtained from Northern Lipids Inc., Canada, while 3,3′-dipropylthiadicarbocyanine iodide (diS-C₃-5), 4-fluoro-7-nitrobenz-2-oxa-1,3-diazole (NBD-fluoride), and tetramethylrhodamine-succinimidyl ester were purchased from Invitrogen, India. Escherichia coli O111:B4 lipopolysaccharide and polymyxin B were obtained from Sigma. For cell culture, RPMI and fetal bovine serum were purchased from Sera Laboratories, West Sussex, United Kingdom. Gibco 100× antibiotic-antimycotic was purchased from Invitrogen Corporation. Sterile polystyrene tissue culture flasks and 96-well plates were procured from Greiner Bio-One, while 6-well plates were from Corning Inc. The rest of the reagents were of analytical grade and were procured locally; buffers were prepared in Milli-Q water (USF-ELGA).

Cell lines and animals.

THP-1 and NIH 3T3 cell lines were obtained from the CSIR-Central Drug Research Institute (CDRI), Lucknow, India, cell line repository. The cell lines were maintained in RPMI medium supplemented with 10% fetal bovine serum and antibiotics in an Innova CO₂ incubator. The animals used for the experiments were provided by the National Laboratory Animal Center, CSIR-CDRI (Lucknow, India). All animal procedures were carried out according to protocols approved by the CSIR-CDRI Animal Ethics Committee (approval no. IAEC/2010/79) and the National Laboratory Animal Centre (Lucknow, India). The animals were properly anesthetized before the experiments, and all treatments were performed in a way that minimized the suffering of the animals. Our animal protocols adhered to the guidelines of the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA; registration no. 34/GO/ReBi/S/99CPCSEA, dated 12 March 15) of the Government of India.

Peptide synthesis, fluorescent labeling, and purification.

Peptide synthesis was done manually utilizing a solid-phase method on rink amide MBHA resin and Fmoc chemistry as described previously (8, 11, 16–20). Labeling of the peptides at their N termini by NBD was achieved by the usual procedure, as described earlier (21). Cleavage of labeled and unlabeled peptides from resins and their purification by reverse-phase HPLC were accomplished as reported previously (21, 22).

Assay of hemolytic and cytotoxic activities of the peptides.

The hemolytic activities of piscidin-1 and its designed analogs against human red blood cells (hRBCs) were measured by assaying their efficacy of lysis of hRBCs, as described elsewhere (11, 23). For this experiment with human blood, we have an approval from the CSIR-CDRI Ethics Committee (approval no. CDRI/IEC/2014/A5).

To examine the cytotoxicity of the peptides, the viability of murine NIH 3T3 cells was determined by a standard 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay as described earlier (17, 21, 24).

Assays of antibacterial activities of the peptides.

Assays of the antibacterial activities of all the peptides against different Gram-positive and Gram-negative bacteria were done in 96-well microtiter plates, as described earlier (10, 21, 23).

Preparation of SUVs.

Small unilamellar vesicles (SUVs) were prepared by a standard procedure by employing a bath-type sonicator (Laboratory Supplies Company, New York, NY) with the required amounts of either PC-cholesterol (8:1, wt/wt) or PC-PG (3:1, wt/wt), as described elsewhere (8, 11).

Assay of peptide-induced dissipation of diffusion potential.

The peptide-induced depolarization of the lipid bilayer was measured by determining the ability of the peptides to dissipate the diffusion potential across lipid vesicles composed of zwitterionic PC-Chol (8:1, wt/wt) or negatively charged PC-PG (3:1 wt/wt) by employing a potential-sensitive dye, diS-C₃-5, as described in earlier reports (16, 22).

Detection of peptide-induced membrane damage of hRBCs and bacterial cells.

Peptide-induced damage of the hRBC membrane was determined by staining the cells (∼3.0 × 10⁷ cells/ml) with Alexa Fluor-annexin V following their treatment with a particular peptide at room temperature for 10 to 15 min and by analyzing it with a Becton Dickinson FACSCalibur flow cytometer using Cell Quest Pro software (25, 26). To examine the peptide-induced damage to the integrity of the E. coli ATCC 25922 cell membrane, mid-log-phase cells were incubated with the peptides for 1 h at 37°C with constant shaking, followed by washing two times with phosphate-buffered saline (PBS), and the cells were then incubated further with propidium iodide (PI) at 4°C for 30 min. The peptide-induced damage to the bacterial cells was analyzed by use of a flow cytometer as mentioned above and reported elsewhere (21).

CD studies.

The circular dichroism (CD) spectra of the peptides in PBS (pH 7.4) and in zwitterionic PC-Chol (8:1, wt/wt) and negatively charged PC-PG (3:1 wt/wt) lipid vesicles were recorded on a Jasco J-815 spectropolarimeter (21, 22).

Localization of peptides on mammalian and bacterial membranes by confocal microscopy.

Fresh hRBCs (4% [vol/vol] in PBS) and E. coli cells (∼10⁶ CFU/ml) were incubated with NBD-labeled peptides in PBS at 37°C with gentle shaking for 30 min and for 1 h in the case of hRBCs and bacterial cells, respectively. The cells were washed with PBS three times and then fixed with 2% paraformaldehyde for 10 to 15 min. After extensive washing with PBS, images of the cells were taken with a Carl Zeiss LSM 510 Meta confocal laser scanning microscope (Carl Zeiss, Jena, Germany). Images were analyzed using Zeiss AIM (version 4.2) software (10).

SEM.

The morphological changes to E. coli induced by the piscidin-1-derived peptides were studied using scanning electron microscopy (SEM), as described earlier (22, 27). Micrographs were taken at magnifications of ×24,000. About 200 cells from two stubs (SEM sample holders) for each sample were analyzed.

Endotoxin neutralization (LAL) assay.

The ability of peptides to bind LPS was assessed using a quantitative chromogenic Limulus amoebocyte lysate (LAL) assay (QCL-1000 kit; catalog number 50-647U; Lonza), as reported earlier (15, 28–30).

Measurement of cytokine expression levels in supernatant.

Enzyme-linked immunosorbent assays (ELISAs) were carried out to estimate the amount of tumor necrosis factor alpha (TNF-α) and interleukin-1β (IL-1β) secreted by LPS-treated THP-1 cells in the presence of peptides after 4 to 6 h of incubation. The levels of these cytokines in the culture supernatant of untreated and LPS-treated cells were taken as the minimum and maximum levels, respectively, to calculate the percent inhibition by the peptides (31). The concentrations of TNF-α and IL-1β in the samples were assessed using enzyme-linked immunosorbent assay kits for human TNF-α (BD Biosciences) and IL-1β (BD Biosciences) according to the manufacturer's protocol. The experiments were repeated thrice, and the average values of the cytokine concentrations determined are included in Results. Similar ELISAs were performed to estimate the amount of TNF-α and IL-6 secreted by BALB/c mouse blood, which was collected from the orbital sinus 4 h after LPS injection, by using enzyme-linked immunosorbent assay kits for mouse TNF-α (BD Biosciences) and IL-6 (BD Biosciences).

In vivo studies and treatment of mice.

Female BALB/c mice (National Laboratory Animal Center, Central Drug Research Institute, Lucknow, India) were provided a standard laboratory diet and water ad libitum and housed under controlled environmental conditions. Each of the mice weighed approximately 25 to 30 g at the start of the experiments. All the mice were divided into experimental groups for LPS and peptide administration (five animals per group). Mice in different experimental groups were treated with 12 mg/kg of body weight E. coli O111:B4 LPS in the presence and in the absence of peptide. The mice treated with saline only, in the absence of LPS, were considered experimental controls. To study the efficacy of a peptide in neutralizing LPS-induced proinflammatory responses in mice at ∼5 min after LPS treatment, each peptide was administered intraperitoneally (i.p.) at a site other than the one used for LPS injection (32, 33). Further, to investigate any cytotoxic effect of the peptides, mice were administered the peptide in the absence of LPS. The survival and physical activity of the experimental groups of mice were recorded at different time points after commencement of the experiments for 7 days per the predetermined scoring criteria, as reported elsewhere (34).

Statistical analysis.

For statistical evaluation, the data were analyzed using Prism software (version 5; GraphPad). For survival analysis, the log rank (Mantel-Cox) test was used to determine statistically significant differences (32, 33).

RESULTS AND DISCUSSION

Design of piscidin-1 analogs.

Piscidin-1, comprised of 22 amino acid residues, contains a very long heptad repeat sequence which is located from amino acids 2 to 19 (Fig. 1A). The a (2nd, 9th, and 16th) positions of the identified heptad repeat sequence are occupied by one phenylalanine and two isoleucine residues, whereas the d positions are occupied by isoleucine, valine, and leucine residues (one amino acid each). To look into the possible role of this important structural motif, six analogs of piscidin-1 were designed by selectively replacing single isoleucine residues located at one d (5th) position and two a (9th and 16th) positions with single alanine as well as valine residues (Fig. 1). The positions of the amino acid substitutions were chosen in such a way that the roles of amino acids located at both the a and d positions of almost the entire length of the heptad repeat were addressed. Both alanine and valine residues at the a or d position are not known for assisting with the self-assembly properties of peptides with a heptad repeat sequence (21, 35). However, since alanine possesses a lower hydrophobicity than isoleucine, replacements with similarly hydrophobic valine residues were also made. The objectives of these substitutions were to assess (i) the contribution of the hydrophobicity of amino acids at these specific positions to the antimicrobial, cytotoxic, and antiendotoxin activities of piscidin-1 and (ii) to determine the effect of replacement of an isoleucine residue at these a/d positions with a similar hydrophobic residue, valine, which is not known for participating in the self-assembling property of a peptide with a heptad repeat sequence (21, 35). The analog I5F,F6A-piscidin-1 was designed to investigate how a phenylalanine residue known for contributing to the self-assembly of a peptide (10, 36) influences the biological property of a peptide when it is placed from a nonheptadic position (6th) to a d (5th) position of the heptad repeat of piscidin-1. To further underscore the distinction between an isoleucine and an equally hydrophobic valine residue, the analog V12I-piscidin-1 was designed by replacing the valine residue at a d (12th) position of the heptad repeat of piscidin-1 with an isoleucine residue, which, unlike valine (21, 35), is known to contribute to the self-assembly of a peptide with a heptad repeat sequence (37, 38). Helical wheel projections of piscidin-1-derived peptides are shown in Fig. 1B and indicate that the amino acid substitutions were made in the hydrophobic face of piscidin-1.

FIG 1.

(A) Sequences and molecular weights of piscidin-1 and its analogs. Substituted amino acids are underlined. The calculated and observed molecular weights of piscidin-1 and its analogs are given in the two columns on the right. (B) Helical wheel projections of piscidin-1 and its analogs. Substituted amino acids are circled in red, and the names of the peptides are incorporated in the corresponding projections.

Replacement of a single isoleucine residue at the a or d position with an alanine residue drastically reduced the hemolytic and cytotoxic activities of piscidin-1, while replacement with a valine residue showed a moderate effect.

The role of a hydrophobic amino acid(s) at different heptadic positions of piscidin-1 in its hemolytic activity against hRBCs and cytotoxic activity against murine NIH 3T3 fibroblasts was examined. Among the piscidin-1-derived peptides, the three single alanine-substituted piscidin-1 analogs were appreciably less hemolytic against hRBCs (Fig. 2A). However, the hemolytic activity of the single valine-substituted piscidin-1 analogs was significantly greater than that of the corresponding single alanine-substituted versions (e.g., I5V-piscidin-1 versus I5A-piscidin-1) (Fig. 2A and Table 1) but was also appreciably lower than that of the native peptide, despite the fact that valine possesses hydrophobicity similar to that of isoleucine. The data suggest a definite role of hydrophobic amino acids at these a and d positions of the heptad repeat of piscidin-1 in determining its hemolytic activity. The results also indicate a difference between isoleucine and valine residues in determining the hemolytic activity of piscidin-1 when they are placed at these specific positions of the heptad repeat of piscidin-1. Even though I5F,F6A-piscidin-1 and I5A-piscidin-1 had the same amino acid composition, I5F,F6A-piscidin-1 to some extent showed higher hemolytic activity than I5A-piscidn-1 (Fig. 2A and Table 1), which indicated the impact of interchanging an alanine residue at a d position of the heptad repeat of piscidin-1 with an adjacent phenylalanine residue in determining its lytic property against hRBCs. The most dramatic result according to hemolytic activity was observed for V12I-piscidin-1, which showed almost 2-fold the hemolytic activity of piscidin-1 after the replacement of a valine residue at a d position of its heptad repeat with an isoleucine residue (Fig. 2A and Table 1). The data undoubtedly showed a stronger impact of an isoleucine residue than a valine residue in determining the hemolytic activity of piscidin-1 when they are placed at the same d position of the heptad repeat sequence of this peptide. The results altogether indicate the role of hydrophobic isoleucine residues positioned at different a and d positions in determining the hemolytic activity of piscidin-1. These results support data previously presented in the literature (8–11, 16). Further, the hemolytic activity of the NBD-labeled versions of these peptides against the hRBCs was assayed. The results indicated that NBD-labeled piscidin-1 and its analogs showed hemolytic activity very similar to that of their unlabeled versions (data not shown), suggesting that labeling of these peptides with NBD at their N termini did not have much of an impact on their cytolytic activities. The cytotoxic activities of these peptides were further examined by looking into the viability of the NIH 3T3 murine fibroblasts by determining the activity of the mitochondrial dehydrogenase in the presence of these peptides by an MTT assay. The cytotoxic activities of piscidin-1 and its analogs, evaluated from the viability of NIH 3T3 cells in their presence, followed the same trend as their hemolytic activities (Fig. 2B) against the hRBCs. The results altogether showed a significant effect of replacement of single isoleucine residues at different a or d positions with an alanine residue on the hemolytic and cytotoxic activities of piscidin-1, whereas replacement of these single isoleucine residues with single valine residues demonstrated a moderate effect on the cytotoxicity of piscidin-1, probably due to its significant hydrophobic character.

FIG 2.

(A and B) Dose-dependent hemolysis of hRBCs (A) and viability of murine NIH 3T3 cells (B) in the presence of piscidin-1 and its analogs. Symbols: black squares, piscidin-1; red circles, I5A-piscidin-1; green up-pointing triangles, I9A-piscidin-1; blue down-pointing triangles, I16A-piscidin-1; cyan diamonds, I5V-piscidin-1; magenta left-pointing triangles, I9V-piscidin-1; yellow right-pointing triangles, I16V-piscidin-1; dark yellow hexagons, I5F,F6A-piscidin-1; navy blue stars, V12I-piscidin-1. Each datum point is the average from three independent experiments, and error bar represents standard deviations. (C) Peptide- induced membrane damage of E. coli ATCC 25922 cells detected by PI staining following treatment with the different peptides. The lower left quadrant of each panel depicts unstained cells, whereas the lower right quadrant depicts stained cells. The concentrations of the peptides were ∼7.5 μM. (D) Peptide-induced damage of the hRBC membrane, detected by Alexa Fluor 488-annexin V staining. The left quadrant of each panel depicts unstained cells, whereas the right quadrant depicts stained cells. The concentrations of the peptides were 25.0 μM. Ten thousand events were recorded for each sample, and the control panels show the negligible PI and Alexa Fluor-annexin V staining of the bacteria and hRBCs, respectively, in the absence of any peptide. SSC, side scatter; PIS, piscidin; LR, lower right quadrant; UR, upper right quadrant.

TABLE 1.

Antibacterial activity of piscidin-1 and its analogs against different bacteria in the absence and presence of 10% serum and their therapeutic indices

Peptide	MIC (μM)									HC50a (μM)	GMb MIC (μM)	Therapeutic indexc
	E. coli ATCC 25922	S. aureus ATCC 25923	P. aeruginosa ATCC BAA-427	K. pneumoniae ATCC 27736	B. subtilis ATCC 6633	In serum		Of labeled peptide
						E. coli ATCC 25922	S. aureus ATCC 25923	E. coli ATCC 25922	S. aureus ATCC 25923
Piscidin-1	3.0 ± 0.4	1.25 ± 0.2	6.0 ± 0.8	3.8 ± 0.6	8.0 ± 1.1	15 ± 2.0	12 ± 2.0	3.5 ± 0.5	1.5 ± 0.5	15	3.68	4.07
I5A-piscidin-1	10.0 ± 1.4	5.0 ± 0.7	20.0 ± 2.5	12.0 ± 1.6	16.0 ± 2.0	>50	>50	12 ± 2.0	6.0 ± 1.0	500	11.39	43.89
I9A-piscidin-1	5.0 ± 0.7	2.5 ± 0.4	10.0 ± 1.4	6.0 ± 0.8	8.0 ± 1.1	20 ± 2.5	18 ± 2.0	6.0 ± 1.0	3.0 ± 0.5	150	5.69	26.36
I16A-piscidin-1	6.0 ± 0.8	2.5 ± 0.4	12.0 ± 1.6	8.0 ± 1.1	8.0 ± 1.1	20 ± 2.5	18 ± 2.0	6.5 ± 1.0	3.0 ± 0.5	500	6.49	77.04
I5V-piscidin-1	5.0 ± 0.7	2.5 ± 0.4	12.0 ± 1.6	6.0 ± 0.8	8.0 ± 1.1	20 ± 2.5	18 ± 2.0	5.5 ± 0.8	3.0 ± 0.5	40	5.90	6.77
I9V-piscidin-1	5.0 ± 0.7	2.0 ± 0.4	12.0 ± 1.6	6.0 ± 0.8	8.0 ± 1.1	20 ± 2.5	18 ± 2.0	5.5 ± 0.8	2.4 ± 0.4	35	5.65	6.19
I16V-piscidin-1	5.0 ± 0.7	2.0 ± 0.4	12.0 ± 1.6	6.0 ± 0.8	8.0 ± 1.1	20 ± 2.5	18 ± 2.0	5.5 ± 0.8	2.4 ± 0.4	40	5.65	7.07
I5F,F6A-piscidin-1	10.0 ± 1.4	5.0 ± 0.7	24.0 ± 3.0	12.0 ± 1.6	16.0 ± 2.0	>50	>50	12 ± 2.0	6.0 ± 1.0	150	11.81	12.70
V12I-piscidin-1	2.0 ± 0.4	1.0 ± 0.2	4.0 ± 0.6	3.0 ± 0.4	6.0 ± 0.8	12 ± 2.0	10 ± 1.4	2.5 ± 0.4	1.0 ± 0.2	7	2.70	2.59

^aHC₅₀, mean concentration of peptide producing 50% hemolysis.

^bGM, geometric mean.

^cThe therapeutic index is HC₅₀/geometric mean MIC.

Piscidin-1 analogs showed antibacterial activity comparable to that of the native peptide.

Piscidin-1 and its analogs were tested for their bacterial growth-inhibiting activity against Gram-positive and Gram-negative bacteria in liquid cultures. In contrast to their hemolytic activity/cytotoxicity, all the single alanine-substituted piscidin-1 analogs showed appreciable antimicrobial activity. However, their activity to some extent was lower than that of the native peptide (Table 1). For example, I9A-piscidin-1 and I16A-piscidin-1 showed activity similar to that of piscidin-1 against the Gram-positive bacterium Bacillus subtilis; however, the MIC value of I5A-piscidin-1 against B. subtilis was twice the MIC value of piscidin-1 against the same bacteria. The MIC values of I9A-piscidin-1 and I16A-piscidin-1 against the Gram-negative bacteria E. coli, Pseudomonas aeruginosa, and Klebsiella pneumoniae and the Gram-positive bacterium Staphylococcus aureus were 2-fold the MIC values of piscidin-1 against the same bacteria. On the other hand, I5A-piscidin-1 showed MIC values ∼3-fold higher than the MIC values of piscidin-1 against E. coli, P. aeruginosa, and K. pneumoniae, whereas its MIC value against S. aureus was about four times that of piscidin-1 (Table 1). Taken together, the data indicate that the replacement of isoleucine residues at the a or d position of the identified heptad repeat of piscidin-1 with alanine residues had only a moderate effect on the antibacterial properties of piscidin-1, while the single valine-substituted piscidin-1 analogs to some extent showed greater antibacterial activity than the single alanine-substituted analogs and their activity was very much comparable to that of the native piscidin-1. The antibacterial activity of I5F,F6A-piscidin-1 was very close to that of its parent peptide, I5A-piscidin-1. Unlike its augmented cytotoxicity, V12I-piscidin-1 showed only a slightly higher level of antibacterial activity than piscidin-1. Thus, irrespective of their cytotoxic properties, all the designed piscidin-1 analogs significantly retained the antibacterial activity of piscidin-1, and variations in their antibacterial activity could be due to differences in their hydrophobicities.

The antibacterial activities of these peptides against a Gram-positive bacterium and a Gram-negative bacterium, namely, S. aureus ATCC 25923 and E. coli ATCC 25922, respectively, were also assayed in the presence of 10% serum (39) (Table 1). Though increases in the MIC values of piscidin-1 and all its analogs were observed in the presence of serum, indicating a reduction of their antibacterial activity, all of the peptides except I5A-piscidin-1 and I5F,F6A-piscidin-1 exhibited appreciable antibacterial activity against both bacteria. Moreover, the antibacterial activity of NBD-labeled piscidin-1 and its analogs against a Gram-negative bacterium, E. coli ATCC 25922, and a Gram-positive bacterium, S. aureus ATCC 25923, was examined. The NBD-labeled peptides exhibited a slight increase in MIC values compared to those of their unlabeled versions; more specifically, each of these labeled peptides retained 80 to 90% of the antibacterial activity of the corresponding unlabeled peptide (Table 1). Thus, the results suggest that even after labeling with NBD, these peptides appreciably retained their antibacterial property, a finding which is consistent with our earlier observations (10) and also indicated the meaningfulness of the confocal microscopic studies performed with these labeled peptides.

Alanine-substituted piscidin-1 analogs showed appreciably enhanced therapeutic indices compared with those of the native peptide and its other analogs.

The therapeutic index of piscidin-1 and its analogs was calculated as the ratio of the mean concentration of peptide producing 50% hemolysis (HC₅₀) from the in vitro data and its geometric mean MIC value (40) against all five bacteria tested. Thus, larger values for the therapeutic index indicate greater antimicrobial specificity (22). As shown in Table 1, the therapeutic indices for piscidin-1, I5A-piscidin-1, I9A-piscidin-1, I16A-piscidin-1, I5V-piscidin-1, I9V-piscidin-1, I16V-piscidin-1, I5F,F6A-piscidin-1, and I12V-piscidin-1 were 4.07, 43.89, 26.36, 77.04, 6.77, 6.19, 7.07, 12.70, and 2.59, respectively. Thus, the results suggest that replacement of a single isoleucine residue with an alanine residue at the a or d position of the identified heptad repeat sequence in piscidin-1 gives a therapeutic index higher than that of the corresponding construct with a single valine substitution at one of these positions, and I16A-piscidin-1 showed the highest therapeutic index among all the piscidin-1-derived peptides studied here.

The literature indicates that there have been efforts to design nontoxic analogs of piscidin-1. For example, glycine residues at positions 8 and 13 were replaced with proline and alanine residues (41). Glycine-to-proline substitutions at the 8th and 13th positions yielded better selectivity in piscidin-1, which was evident by relative selectivity indices of 13.22 and 6.91 for the corresponding piscidin-1 analogs, whereas replacement of the glycine residues at the 8th and 13th positions with alanine residues yielded selectivity indices of 0.48 and 0.61 for the resulting analogs, indicating that the cell selectivity of these analogs is lower than that of piscidin-1 (41). The piscidin-1 analog in which the glycine at the 8th position was replaced by Nlys (Lys peptoid residue) showed antibacterial activity comparable to that of piscidin-1 and cytotoxic properties significantly reduced in comparison to those of piscidin-1 (42), resulting in its higher therapeutic index (therapeutic index, 140) compared with that of the native peptide (therapeutic index, 1.6). The role of the phenylalanine residues near the N terminus of piscidin-1 and the valine residue at the 10th position has recently been addressed. It has been shown that a phenylalanine residue at the 2nd position plays a crucial role in both the cytotoxic and the antibacterial properties of piscidin-1 (5). Replacement of this phenylalanine residue with alanine and lysine residues enhanced the cell selectivity of piscidin-1, as evidenced by relative selectivity indices of 22 and 51 for the corresponding piscidin-1 analogs, respectively. However, so far no structural element in piscidin-1 that shows a strong impact on the overall structure, function, and biological properties of piscidin-1 has been identified.

Piscidin-1 and all its analogs significantly damaged the bacterial membrane, whereas differences in causing damage to the hRBC membrane were observed among these peptides.

To investigate whether piscidin-1 and its analogs can alter the membrane organization of bacteria and mammalian cells, PI staining of E. coli ATCC 25922 cells and Alexa Fluor-annexin V staining of hRBCs following their treatment with these peptides were carried out. PI staining of bacteria is an indication of the binding of PI to bacterial nucleic acids, which is possible only after the bacterial membrane is damaged. All the peptides, including piscidin-1 and its analogs, induced extensive staining of E. coli (Fig. 2C), suggesting the ability of these peptides to damage the membrane organization of the bacteria. E. coli bacteria treated with V12I-piscidin-1 showed the highest level of staining with PI, followed by bacteria treated with piscidin-1, I9V-piscidin-1, I16V-piscidin-1, I5V-piscidin-1, I16A-piscidin-1, and I9A-piscidin-1, in that order, whereas bacteria treated with I5A-piscidin-1 and I5F,F6A-piscidin-1 to some extent showed lower levels of staining with PI than the bacteria treated with the other piscidin-1-derived peptides, reflecting the relatively lower antibacterial activities of these peptides. The membrane-impermeant dye annexin V-Alexa Fluor specifically binds to phosphatidylserine (PS), which is found in the inner leaflet of the hRBC/mammalian cell membrane. Annexin V-Alexa Fluor can access the inner leaflet of hRBCs when their membranes are damaged. Thus, peptide-induced annexin V-Alexa Fluor staining of hRBCs is directly related to the membrane-damaging property of that peptide. Annexin V-Alexa Fluor staining of hRBCs suggested that V12I-piscidin-1, piscidin-1, and the three single valine-substituted piscidin-1 analogs significantly damaged the organization of the hRBC membrane, while the three single alanine-substituted piscidin-1 analogs were appreciably inactive in this regard, suggesting that these peptides have hemolytic activities lower than those of the former group peptides. Previous studies also reported the hemolytic activity of piscidin-1 against human RBCs as well as its cytotoxicity toward a human cell line(s) (5), which are findings consistent with those of the present study (Fig. 2D).

Differences among the piscidin-1-derived peptides in permeabilizing mammalian membrane-mimetic, zwitterionic lipid vesicles but not bacterial membrane-mimetic, negatively charged lipid vesicles.

The majority of the antimicrobial peptides primarily target the membranes of microorganisms to induce their antibacterial activity. Therefore, to understand the basis of the antibacterial and cytotoxic activity of piscidin-1 and its designed analogs, the permeabilization of bacterial membrane-mimetic, negatively charged lipid vesicles and mammalian membrane-mimetic, zwitterionic lipid vesicles was studied in the presence of these peptides (Fig. 3). Peptide-induced permeabilization of different kinds of lipid vesicles was determined by examining the ability of the peptides to dissipate the diffusion potential across the phospholipid vesicles, which was detected by an increase in the fluorescence of a potential-sensitive dye, diS-C₃-5.

FIG 3.

(A and B) Plots of fluorescence recovery, which is a measure of peptide-induced membrane permeabilization, versus peptide concentration in bacterial and mammalian membrane-mimetic PC-PG (3:1, wt/wt) (A) and PC-Chol (8:1, wt/wt) (B) lipid vesicles. Symbols: black squares, piscidin-1; red circles, I5A-piscidin-1; green up-pointing triangles, I9A-piscidin-1; blue down-pointing triangles, I16A-piscidin-1; cyan diamonds, I5V-piscidin-1; magenta left-pointing triangles, I9V-piscidin-1; yellow right-pointing triangles, I16V-piscidin-1; dark yellow hexagons, I5F,F6A-piscidin-1; and navy blue stars, V12I-piscidin-1. (C) Scanning electron micrographs of E. coli ATCC 25922 in the absence and presence of different peptides, showing bacteria without any treatment (Control) and after treatment with 10-fold the MIC values of piscidin-1 and its analogs, as marked at the bottom of each image.

Piscidin-1 and its analogs induced appreciable permeabilization of bacterial membrane-mimetic PC-PG lipid vesicles, as measured by the percentage of fluorescence recovery (Fig. 3A). However, while V12I-piscidin-1 and piscidin-1 to some extent exhibited higher permeabilization efficiencies, I5A-piscidin-1 and I5F,F6A-piscidin-1 showed less activity than the other piscidin-1-derived peptides in permeabilizing PC-PG lipid vesicles (Fig. 3A). Nevertheless, the most cytotoxic peptide, V12I-piscidin-1, induced the highest permeabilization in mammalian membrane-mimetic PC-Chol lipid vesicles; this was followed by the cytotoxic, native piscidin-1 and its single valine-substituted piscidin-1 analogs. I5F,F6A-piscidin-1 and I9A-piscidin-1 also permeabilized PC-Chol vesicles but did so at a level significantly lower than that of piscidin-1 and its valine-substituted analogs. However, I16A-piscidin-1 and I5A-piscidin-1 showed the lowest permeabilization in this kind of lipid vesicle (Fig. 3B). Thus, the results obtained from the peptide-induced damage of the membrane organization of bacterial and mammalian cells (Fig. 2) and permeabilization of bacterial and mammalian membrane-mimetic lipid vesicles (Fig. 3) matched the antibacterial and hemolytic/cytotoxic activities of piscidin-1 and its analogs, respectively.

Visualization of bacterial morphology after treatment with piscidin-1 and its analogs by SEM.

To get more insight into the mode of action of piscidin-1 and its analogs, the morphology of E. coli ATCC 25922 cells was visualized by scanning electron microscopy (SEM) following treatment with these peptides at 10-fold the MIC for 60 min. Bacteria not treated with any of these peptides appeared to have a smooth surface by SEM (Fig. 3C). E. coli cells treated with either piscidin-1, V12I-piscidin-1, any of the single valine-substituted analogs, or the single alanine-substituted analog I16A-piscidin-1 showed prominent changes in cellular morphology, including wrinkling, surface roughening, membrane blebbing, leakage of bacterial contents, and lysis of the bacterial cell. The analog I9A-piscidin-1 had a moderate effect on bacterial morphology (Fig. 3C), while among these piscidin-1-derived peptides, I5A-piscidin-1 and I5F,F6A-piscidin-1 caused relatively less damage to the bacterial morphology.

Piscidin-1 and its analogs exhibited comparable secondary structures in negatively charged phospholipid membranes, but appreciable differences between them were observed in zwitterionic phospholipid membranes.

To assess the secondary structures of piscidin-1 and its analogs, circular dichroism studies of these peptides were performed in the presence of PBS (pH 7.4), bacterial membrane-mimetic, negatively charged lipid vesicles (PC-PG, 3:1 wt/wt), and mammalian membrane-mimetic, zwitterionic (PC-Chol, 8:1, wt/wt) lipid vesicles. Only piscidin-1 and V12I-piscidin-1 and not the other piscidin-1 analogs to some extent adopted helical structures in PBS (spectra not shown). However, the CD spectra suggested that in the presence of PC-PG lipid vesicles, piscidin-1 and all its analogs adopted significant and quite comparable helical structures (Fig. 4A). In the presence of PC-Chol lipid vesicles, piscidin-1 maintained its high helical content; the CD spectra suggested that the most cytotoxic peptide, V12I-piscidin-1, showed the highest helical content among these peptides, and all the valine-substituted analogs also exhibited appreciable helical contents, but their helical contents were less than that of the native peptide. However, all three alanine-substituted analogs exhibited lower helical contents than their valine-substituted counterparts in PC-Chol lipid vesicles. Similar to their relative cytotoxic properties, I5F,F6A-piscidin-1 to some extent had a higher helical content than its parent peptide, I5A-piscidin-1, in PC-Chol lipid vesicles (Fig. 4B).

FIG 4.

Determination of secondary structures of piscidin-1 and its designed analogs in the presence of negatively charged PC-PG lipid vesicles (A) and zwitterionic PC-Chol lipid vesicles (B). The concentration of each kind of lipid vesicle was ∼375 μM, and the concentration of each of the peptides was ∼25 μM. Symbols: Black squares, piscidin-1; red circles, I5A-piscidin-1; green up-pointing triangles, I9A-piscidin-1; blue down-pointing triangles, I16A-piscidin-1; cyan diamonds, I5V-piscidin-1; magenta left-pointing triangles, I9V-piscidin-1; yellow right-pointing triangles, I16V-piscidin-1; dark yellow hexagons, I5F,F6A-piscidin-1; and navy blue stars, V12I-piscidin-1.

Localization of piscidin-1 and its analogs on bacteria and hRBCs by confocal microscopy.

The cellular localization of piscidin-1 and its analogs, including the three single alanine-substituted analogs, two single valine-substituted analogs, I9V-piscidin-1, I16V-piscidin-1, and V12I-piscidin-1, on E. coli ATCC 25922 cells and hRBCs was probed by confocal microscopy by employing the NBD-labeled versions of these peptides. As was evident from the confocal microscopic images, NBD-labeled piscidin-1 and its analogs mostly accumulated on the bacterial membrane (Fig. 5A). The results suggested the similar localization of NBD-labeled piscidin-1 and its various analogs on E. coli cells.

FIG 5.

Confocal microscopy for studying the localization of NBD-labeled peptides on bacteria and hRBCs. (A) Localization of NBD-labeled piscidin-1 and its analogs (∼5.0 μM) on E. coli ATCC 25922 cells; (B) localization of NBD-labeled piscidin-1 and its analogs (∼10.0 μM) on hRBCs.

On the other hand, NBD-labeled piscidin-1, V12I-piscidin-1, I9V-piscidin-1, and I16V-piscidin-1 localized effectively on the hRBCs, as seen by the prominent green fluorescence on these cell membranes (Fig. 5B). Differential interference contrast (DIC) images showed that NBD-labeled piscidin-1, I9V-piscidin-1, I16V-piscidin-1, and V12I-piscidin-1 inflicted significant damage on the integrity of hRBCs. Although the cells were visible in the DIC panels, the fluorescence signals of NBD-labeled I5A-piscidin-1, I9A-piscidin-1, and I16A-piscidin-1 on these cells were too weak to be visualized, which implied their weak binding to hRBCs. The localization of NBD-I5V-piscidin-1 on E. coli cells and hRBCs was quite similar to that of the other two NBD-labeled single valine-substituted piscidin-1 analogs, whereas the localization of NBD-I5F,F6A-piscidin-1 on E. coli cells was comparable to that of NBD-I5A-piscidin-1 and its localization on hRBCs was close to that of any of the NBD-labeled single alanine-substituted piscidin-1 analogs. Therefore, confocal microscopic images of NBD-I5V-piscidin-1 and NBD-I5F,F6A-piscidin-1 on bacteria or hRBCs are not presented.

Piscidin-1 and its analogs neutralized LPS-induced proinflammatory cytokine production in THP-1 cells.

To investigate the consequence of the amino acid substitutions at the a or d position of the heptad repeat sequences of piscidin-1 on its antiendotoxin properties, human THP-1 monocytic cells were stimulated with LPS in the absence and presence of piscidin-1 and its analogs. The level of production of TNF-α and IL-1β observed in LPS-stimulated THP-1 cells was significant compared to that observed in the control or unstimulated cells (see Fig. S1A in the supplemental material). However, at concentration ranging from 7.5 to 15 μM, V12I-piscidin-1, piscidin-1, the three single valine-substituted piscidin-1 analogs, and the single alanine-substituted analogs I9A-piscidin-1 and I16A-piscidin-1 significantly inhibited the production of these two cytokines. I5A-piscidin-1 to some extent was less active than the other two single alanine-substituted piscidin-1 analogs in inhibiting the production of these two cytokines in LPS-stimulated THP-1 cells. I5F,F6A-piscidin-1, which had the same composition as I5A-piscidin-1 but with the amino acids at positions 5 and 6 interchanged with each other, to some extent showed higher activity than its parent peptide (Fig. 6A and B).

FIG 6.

(A and B) Percent inhibition of levels of LPS-induced secretion of TNF-α (A) and IL-1β (B) in the presence of piscidin-1 and its analogs determined by ELISA. (C) Dose-dependent LPS neutralization by piscidin-1 and its analogs determined by the LAL assay. The results are representative of three independent experiments.

An overlap of the structural requirement for the cytotoxic and antiendotoxin properties of melittin (15) was demonstrated in earlier studies. A linear correlation between the cytotoxic and antiendotoxin properties of AMPs with heptad repeat sequences was observed (21). In the present study, all the single valine-substituted piscidin-1 analogs and I5F,F6A-piscidin-1, which had higher levels of cytotoxicity than the corresponding alanine-substituted piscidin-1 analogs, also showed superior antiendotoxin properties (Fig. 6; see also Fig. S1A in the supplemental material). V12I-piscidin-1 demonstrated the highest antiendotoxin activity among all these peptides (Fig. 6; see also Fig. S1A in the supplemental material), and it also exhibited the highest hemolytic activity among all these peptides (Fig. 2 and Table 1). Data on the production of TNF-α and IL-1β in LPS-stimulated THP-1 cells in the absence and presence of the peptides are shown in Fig. S1A in the supplemental material.

LPS-neutralizing activity of the piscidin-1 and its analogs.

The Limulus amoebocyte lysate (LAL) test is generally accepted to indicate the ability of a molecule to neutralize or inhibit LPS (28, 43). Thus, to assess the effect of replacement of the amino acids at the a or d position of the heptad repeat sequence of piscidin-1, the dose-dependent ability of piscidin-1 and its analogs to neutralize LPS was determined by a chromogenic LAL assay (Fig. 6C). V12I-piscidin-1, piscidin-1, I5V-piscidin-1, I9V-piscidin-1, I16V-piscidin-1, I9A-piscidin-1, I16A-piscidin-1, and I5F,F6A-piscidin-1 showed significant and comparable levels of binding to LPS, as evidenced by the substantial inhibition of activation of the LAL enzyme which occurs in the presence of LPS (Fig. 6C). However, I5A-piscidin-1 to some extent was less active than the other piscidin-1-derived peptides.

Treatment of the piscidin-1 analog I16A-piscidin-1 facilitated the survival of LPS-treated mice.

Among the peptides that included piscidin-1 and the piscidin-1-derived analogs, I16A-piscidin-1 was chosen for use in the investigation of antiendotoxin activity in vivo because of its significant antiendotoxin properties, because it had the lowest cytotoxicity, and because it had the highest therapeutic index (Fig. 2 and 6 and Table 1). The experiments were performed as reported earlier (30, 44). Despite their significant antiendotoxin activities, due to their highly cytotoxic natures and lower therapeutic indices, V12I-piscidin-1, native piscidin-1, and the three single valine-substituted analogs of piscidin-1 were excluded from the in vivo assay of survival of mice against LPS challenge (45, 46). LPS-induced toxemia was followed by recording the severity of its symptoms (34) in mice after they were treated with LPS (12 mg/kg) in the absence as well as in the presence of different doses of I16A-piscidin-1 in a 7-day experiment. As is evident from the data in Table 2, symptoms of LPS-induced severe toxemia were observed in mice within 4 h of treatment with LPS alone, and with time the severity of the symptoms worsened (Table 2). Finally, the death of all mice (100%) in this group was recorded within 48 h of LPS treatment (Table 2). However, when LPS-treated mice were further administered a single dose of 1 mg/kg of I16A-piscidin-1, the severity of the symptoms of LPS-induced toxemia was reduced and the majority of the mice slowly recovered from these symptoms (Table 2). Finally, an 80% rate of survival was noticed for this group of mice. Mice treated with a single dose of 2 mg/kg of I16A-piscidin-1 exhibited a better and faster recovery from LPS-induced toxemia than mice treated with the lower dose, and ultimately, 100% recovery of LPS-treated mice treated with a single dose of 2 mg/kg of I16A-piscidin-1 was recorded (Fig. 7A). To look into the in vivo cytotoxicity of I16A-piscidin-1, mice were treated with a single dose of 5 mg/kg of I16A-piscidin-1, and their physical activity along with that of LPS-treated mice treated or not treated with the peptide was recorded. As is evident from the data in Table 2, I16A-piscidin-1 (5 mg/kg) had no cytotoxic effect on mice treated with this peptide only, and 100% survival of this group of mice was recorded at the end of the experiment (Fig. 7A). The mice in the control group, which were treated with saline only, were very active throughout the experimental period, and 100% survival was recorded for that group of mice.

TABLE 2.

Survival and physical activity of different groups of BALB/c mice at different time points after commencement of the experiments

Treatment	Physical activity of each of the five animals ata:
	4 h	8 h	12 h	24 h	48 h	72 h	120 h	168 h
Saline	11111	11111	11111	11111	11111	11111	11111	11111
LPS (12 mg/kg)	33333	33444	33444	55†††	†††††	†††††	†††††	†††††
LPS + I16A-piscidin-1 (1 mg/kg)	33333	22333	22223	22224	1112†	1111†	1111†	1111†
LPS + I16A-piscidin-1 (2 mg/kg)	22333	22222	22222	11111	11111	11111	11111	11111
I16A-piscidin-1 (5 mg/kg)	11111	11111	11111	11111	11111	11111	11111	11111

^aGroups of animals (n = 5 each) were inoculated intraperitoneally with LPS (12 mg/kg) alone, LPS (12 mg/kg) and 1 mg/kg or 2 mg/kg of I16A-piscidin-1, or 5 mg/kg I16A-piscidin-1 or saline alone. The data indicate the physical activity of each of the five animals at the indicated time point rated according to the following criteria: 1, very active; 2, active; 3, less active; 4, slow; 5, lethargic; †, dead.

FIG 7.

Effects of I16A-piscidin-1 against LPS challenge in vivo in mice. (A) Septic shock in BALB/c mice was induced by i.p. injection of E. coli LPS (12 mg/kg) followed by i.p. injection of a single dose of I16A-piscidin-1 (1 mg/kg or 2 mg/kg) or saline only ∼5 min later. The cytotoxicity of I16A-piscidin-1 was examined by i.p. injection of a single dose of 5 mg/kg of I16A-piscidin-1 in mice, and the mouse group treated with saline only was used as a negative control. The survival of the animals (n = 5) was monitored for 7 days, and the P value was determined by the log-rank test. (B) Determination of percent inhibition of TNF-α and IL-6 production in mouse serum 4 h after LPS and I16A-piscidin-1 injection.

Further, we investigated the changes in the levels of the proinflammatory cytokines TNF-α and IL-6 in the serum of BALB/c mice 4 h after LPS administration (12 mg/kg i.p.). When the group of mice receiving LPS (12 mg/kg) was further treated with I16A-piscidin-1 (1 mg/kg and 2 mg/kg), appreciable inhibition of the TNF-α and IL-6 levels in their serum was observed (Fig. 7B). The corresponding TNF-α and IL-6 levels in the serum of mice treated with LPS only and mice treated with both LPS and one of two doses of I16A-piscidin-1 are shown in Fig. S1B in the supplemental material. Thus, the survival data for mice treated with I16A-piscidin-1 only and mice administered both LPS and I16A-piscidin-1 (Fig. 7A) demonstrated the negligible cytotoxicity and in vivo efficacy of I16A-piscidin-1 for the rescue of LPS-treated mice, respectively.

The role of a leucine zipper-like sequence in controlling both the cytotoxic and antiendotoxin properties of melittin (15) was reported earlier. The roles of other kinds of heptad repeat sequences in the hemolytic/cytotoxic properties of antimicrobial peptides have been characterized previously (15, 21). Even a single amino acid substitution at the a or d position of a heptad repeat sequence can significantly reduce the cytotoxicity of an antimicrobial peptide. However, the extent to which an amino acid substitution at the a and/or d position of a heptad repeat sequence of an antimicrobial peptide affects its antiendotoxin properties has not yet been demonstrated that well. An effort to reduce the cytotoxicity of an antimicrobial peptide has also, at times, resulted in the compromise of its antiendotoxin properties (15, 21). Therefore, in the current study, an attempt was made to reduce the cytotoxicity of piscidin-1 by incorporating only single amino acid substitutions in its identified heptad repeat sequence so that there could be a chance that these single alanine-substituted piscidin-1 analogs also appreciably retained the antiendotoxin properties of their native peptide. The data presented here clearly showed that our attempt to create nontoxic piscidin-1 analogs that also possess significant antiendotoxin activity was successful. We presume that due to the length of the identified heptad repeat sequence, which makes up 80% of the length (18 amino acids/22 amino acids) of piscidin-1, incorporation of only a single alanine substitution did not significantly alter its LPS binding, and hence, the single alanine-substituted piscidin-1 analogs significantly retained the antiendotoxin properties of piscidin-1. The relative efficacy of piscidin-1 and its analogs in neutralizing the LPS-induced production of proinflammatory cytokines matched the extent of inhibition of the LPS-induced activation of the LAL enzyme in their presence. The result also probably suggests a crucial role of the interaction of piscidin-1 or its analogs with LPS in inhibiting LPS-induced cytokine production, as has been proposed for a number of antimicrobial peptides (15, 30). The reason for the low level of antiendotoxin activity of I5A-piscidin-1 is not clear at present. We speculate that the isoleucine residue at the 5th position could have a stronger role in piscidin-1 binding to LPS, and therefore, its replacement by alanine probably had a stronger impact on both binding to LPS and the antiendotoxin activity of the native peptide.

The results of this study suggest a crucial role of the hydrophobicity of the amino acid residues at the a and d positions of the heptad repeat of piscidin-1 in determining its hemolytic and antiendotoxin properties. An isoleucine residue at the d position of the heptad repeat of piscidin-1 contributed more prominently to the hemolytic and antiendotoxin properties of piscidin-1 than a valine residue at the same position, despite the similar hydrophobicities of the two amino acids. The data show the difference between the valine and isoleucine residues when they are placed at the same d position of the heptad repeat sequence of piscidin-1. The results also provide a plausible basis for the design of piscidin-1 analogs with antiendotoxin properties but significantly reduced hemolytic/cytotoxic activities by replacement of an isoleucine residue at an a or a d position of the identified heptad repeat with an alanine residue. Taken together, the results show the identification and characterization of a long heptad repeat sequence located in the region from amino acids 2 to 19 of piscidin-1 that has not previously been reported, to our knowledge. We have demonstrated a very simple way to design cell-selective analogs of piscidin-1 with very significant antiendotoxin properties by incorporating single amino acid substitutions at the a/d positions of its heptad repeat sequence. One of these designed analogs also showed considerable antiendotoxin properties in vivo in mice. The present study could be utilized for the design of nontoxic analogs of AMPs with antiendotoxin properties.