Novel D-form of hybrid peptide (D-AP19) rapidly kills Acinetobacter baumannii while tolerating proteolytic enzymes

Antimicrobial peptides (AMPs) are being developed as potent alternative treatments to conventional antibiotics which are unlikely to induce bacterial resistance. They can be designed and modified to possess several druggable properties. We report herein a novel hybrid peptide of modified aurein (A3) and cathelicidin (P7), or A3P7, by a flipping technique. It exhibited potent antibacterial activity against both Gram-negative and -positive pathogenic bacteria but had moderate hemolytic activity. To reduce the sequence length and toxicity, C-terminal truncation was serially performed and eight truncated derivatives (AP12–AP19) were obtained. They had significantly less hemolytic activity while preserving antibacterial activity. Secondary structures of the candidate peptides in environments simulating bacterial membranes (30 mM SDS and 50% TFE), determined by CD spectroscopy, showed α-helical structures consistent with predicted in silico 3D structural models. Among the peptides, AP19 demonstrated the best combination of broad-spectrum antibacterial activity (including toward Acinetobacter baumannii) and minimal hemolytic and cytotoxic activities. A D-form peptide (D-AP19), in which all L-enantiomers were substituted with the D-enantiomers, maintained antibacterial activity in the presence of pepsin, trypsin, proteinase K and human plasma. Both isomers exhibited potent antibacterial activity against multi-drug (MDR) and extensively-drug resistant (XDR) clinical isolates of A. baumannii comparable to the traditional antibiotic, meropenem. D-AP19 displayed rapid killing via membrane disruption and leakage of intracellular contents. Additionally, it showed a low tendency to induce bacterial resistance. Our work suggested that D-AP19 could be further optimized and developed as a novel compound potentially for fighting against MDR or XDR A. baumannii.

Bacterial strains and growth conditions. Ten strains of both Gram-positive and -negative bacteria, including Staphylococcus aureus ATCC 25,923, S. epidermidis ATCC 12,228, Bacillus cereus ATCC 11,778, Listeria monocytogenes 10403S, Salmonella typhimurium ATCC 13,311, Pseudomonas aeruginosa ATCC 27,853, Shigella sonnei ATCC 11,060, A. baumannii ATCC 19,606, Escherichia coli ATCC 25,922 and E. coli O157:H7, were used in this study. Eight clinical isolates of multidrug-resistant (MDR) and extensively drug-resistant (XDR) A. baumannii, isolated from human (blood, sputum, respiratory sample, sterile site sample and urine), were kindly provided by Asst. Prof. Dr. Sakawrat Kanthawong (Faculty of Medicine, Khon Kaen University, Thailand) (Table S1). All bacterial strains, excluding L. monocytogenes 10403S and E. coli, were cultured in tryptic soy broth (TSB). L. monocytogenes was grown in TSB supplemented with 0.6% yeast extract (TSB-YE), and E. coli was cultured in Luria broth (LB). All of bacteria were cultured at 37 °C with continuous shaking at 220 rpm. www.nature.com/scientificreports/ Antimicrobial assays. Minimum inhibitory concentration (MIC) of peptides and antibiotics was assessed by a modified version of the broth microdilution method from the Clinical and Laboratory Standards Institute (CLSI) as previously described 19 . In brief, mid-log phase growths of microbial strains were cultured in Müeller-Hinton broth (MHB) with an initial inoculum of 2-8 × 10 5 CFU/mL. Bacteria suspensions were incubated in sterile 96-well plates with test agents at desired concentrations. MIC was defined as the lowest concentration of peptide that prevented visible growth of bacteria after 24 h incubation with continuous shaking (220 rpm) at 37 °C. Minimum bactericidal concentration (MBC) determination was performed using a modified version of the colony count assay 19 . In brief, fifty µl of all non-turbid wells from MIC experiments were spread on agar plates. The MBC value was defined as the lowest concentration of peptides with no colony growth on the plate. All assays were performed in triplicate.
Hemolysis assay. The hemolytic activity of peptides was determined as the amount of hemoglobin released from lysed human red blood cells (hRBCs) after treatment with peptide 20 . Fresh hRBCs were collected from a healthy volunteer in polycarbonate tubes containing heparin. Then, the collected packed RBCs were washed at least three times (or until the supernatant was clear) with sterile phosphate-buffered saline pH 7.4 (PBS) and centrifuged at 2,000 × g for 5 min. The 2% (v/v) of washed hRBCs in PBS were incubated with serially diluted peptides (0.98 to 250 μg/ml) for 1 h at 37 °C. After centrifugation, the supernatants were monitored for optical density (OD) at 405 nm using a Multiskan FC Microplate Reader 21 . Values for 0% (negative control) and 100% (positive control) lysis were determined by incubating the hRBCs with PBS only (OD Blank ) and 0.1% (v/v) Triton X-100 (OD Triton X-100 ), respectively. The experiments were performed in triplicate. The percent hemolysis was calculated according to the following equation: Experiments associated with human volunteers were carried out in accord with the ethical standards of and approved by the Ethics Committee of Thammasat University (COA No. 066/2562). The informed consent was voluntarily written by all individual participants.
In vitro cytotoxicity assay. The colorimetric 3-(4,5 dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Invitrogen) dye reduction assay was used to determine the cytotoxicity of peptides on mouse fibroblast (L929) cells (NCTC clone 929) according to a modified MTT assay 22 . L929 cells were cultured in Dulbecco's Modified Eagle Medium (DMEM; Gibco) containing 10% fetal bovine serum (FBS; Gibco) and 100 U/ml penicillin-streptomycin in a fully humidified atmosphere of 95% air and 5% CO 2 at 37 °C. The cells (10 5 cells/well) were seeded on 96-well plates and incubated with serially diluted peptide (0.98 to 250 μg/ml) for 24 h. At the end of incubation period, MTT solution (100 µl, 0.4 mg/ml) was added to each well and incubated for 4 h. The supernatants were removed and replaced by 100 µl of DMSO to dissolve the purple formazan crystals. Absorbance was measured using a MultiskanTM FC microplate reader at a wavelength of 570 nm. Cells without peptides served as negative controls. Cell viability (percentage) was calculated using the following equation: Assessment of peptide stability in different environments. The MIC and MBC assays were performed after peptide was exposed to various conditions, including proteolytic enzymes, serum salts and human plasma 23 . AP19 and D-AP19 at final concentration ranging from 0.98 to 250 µg/ml were pre-incubated with a 1 mg/ml final concentration of proteolytic enzyme (trypsin, pepsin and proteinase K) at 37 °C for 1 h. To determine peptide sensitivity to physiological salts, approximately 5 × 10 5 CFU/ml of A. baumannii ATCC 19,606 was mixed with the MHB containing different physiological salts at final concentration as follows: 150 mM NaCl, 4.5 mM KCl, 1 mM MgCl 2 , 6 µM NH 4 Cl, 8 µM ZnCl 2 , 4 µM FeCl 3 and 2.5 µM CaCl 2 . After treatment, the MIC was evaluated as described in 2.4. To test the stability of peptides in human plasma, peptide was mixed with pure human plasma, and the mixture was incubated for 1 h at 37 °C. Then, the mixture was two-fold serially diluted with sterile MHB. The peptides at final concentrations of 0.98 to 250 µg/ml were incubated with an equal volume of bacterial suspension at 37 °C for 24 h. MIC values were determined as described in 2.4.
Time-kill kinetic assays. The kinetics of the bactericidal activity of D-AP19 against A. baumannii ATCC 19,606 were determined by assessment of the time course of bacterial killing. A bacterial inoculum at approximately 5 × 10 5 CFU/ml suspended in MHB was incubated with D-AP19 at a concentration of 7.81 µg/ml (both MIC and MBC) at 37 °C with continuous shaking at 220 rpm. Samples were taken from the 96-well plates at specific time intervals (0, 0.25, 0.5, 1, 2, 4, 6, 8, 10, 12 and 24 h) and tenfold serially diluted in PBS. Aliquots of each dilution were then plated on tryptic soy agar (TSA). Colonies of bacteria were counted after overnight incubation. A control of bacterial growth (no peptide added) was included in each run.
Circular dichroism analysis. The secondary structure of selected designed peptides dissolved in PBS was determined using circular dichroism (CD) spectra on a Jasco-815 spectropolarimeter under nitrogen at 25 °C, using a 0.1-cm-path-length rectangular quartz cell 24 . Peptide spectra were recorded in 3 different environments: PBS, 30 mM sodium dodecyl sulfate (SDS) micelles in PBS, and 50% (v/v) TFE (2,2,2-trifluoroethanol) in PBS. The SDS micelles simulate the anionic amphipathic environment of bacterial membranes, i.e., external negatively charged surface, but hydrophobic internal environment like the chains of phospholipids. The fluorinated %Hemolysis = OD Sample − OD Blank / OD Triton X−100 −OD Blank × 100 % Cell viability = OD 570 of treated sample − OD Blank /(OD 570 of control − OD Blank ) × 100% www.nature.com/scientificreports/ environment poorly interacts with the molecule, and therefore promotes intra-molecular interactions, that often end in a forced structuration (e.g., in alpha-helix) of the peptide. Spectra were determined in triplicate using a 190-260 nm range at a scanning speed of 10 nm/min. After that, the acquired CD signal spectra were converted to mean residue ellipticity using the following equation: where θ M is residue ellipticity (deg. M-1 m-1), θ obs is detected ellipticity adjusted for buffer at a given wavelength (mdeg), M RW is residue molecular weight (M W /number of amino acids), c is peptide concentration (mg/mL), and l is path length (cm). In addition, the CD spectra were analyzed by CDPro software to estimate the content of secondary structures. The CONTIN/LL from the software package was used to analyze the data.
Flow cytometry analysis. Damage to bacterial cell membranes after interaction with peptide was evaluated using flow cytometry with incorporation of fluorescent dyes 25 . Mid-log phase growth of A. baumannii ATCC 19,606 at OD 620 0.05 in MHB was treated with 0.5 × MIC and 1 × MIC of D-AP19, followed by incubation at 37 °C for 0 h, 0.25 h, 1 h or 2 h with continuous shaking at 220 rpm. The treated bacterial cells were centrifuged at 10,000 × g for 10 min. Then, they were washed again to remove unbound-peptide molecules. Each sample was stained with PI (propidium iodide) or BOX (bis-(1,3-dibutylbarbituric acid) trimethine oxonol) fluorescent dye. All data were recorded using a flow cytometer (CytoFlex, Beckman Coulter), counting 25,000 cells in each sample, at a laser excitation wavelength of 488 nm. Forward scatter (FS) and side scatter (SS) indicated cell size and granularity (complexity), respectively. Red (585/342 nm) and green (530/30 nm) fluorescent signals from PI and BOX, respectively, were investigated. The data were analyzed with Kaluza software version 2.1 (Beckman Coulter, Brea, CA, United States). Negative and positive controls were untreated bacterial cells at 0 h and bacterial cells heated for 30 min at 70 °C, respectively. Three independent experiments were performed.

Scanning electron microscopy.
Morphologic changes of bacterial cell surfaces after D-AP19 treatment were investigated by scanning electron microscopy (SEM) as previously described 26,27 . Bacterial cells in midlog phase were diluted with PBS to obtain an OD 620 of 0.05 and incubated at 37 °C for 0.25 h with peptide at a concentration of 0.5 × MIC. After incubation, the treated bacterial cells were centrifuged at 10,000 g for 10 min and washed 3 times with PBS, pH 7.2. Then, they were filtered through 0.22 µM mixed cellulose ester (MCE) membrane filters to retain the treated bacterial cells. Samples were pre-fixed by immersion into 2.5% (vol/vol) glutaraldehyde-PBS, then post-fixed with 1% OsO 4 in DW. After washing 3 times, they were dehydrated using a graded ethanol series of 20%, 40%, 60%, 80% and 100%, 15 min in each dilution. After that, the samples were transferred into absolute ethanol 2 times, 15 min each time. Specimens were then dried and coated by platinum particle using Sputter Coater (Quorum Q150R ES; Quorum). Processed bacterial cells were observed using a scanning electron microscope (Hitachi SU8020; Hitachi, Japan).
Transmission electron microscopy. The structural changes and integrity of bacterial membranes were investigated by transmission electron microscopy (TEM) as previously described 28 . Treatment of the bacterial samples was conducted in the same manner as described for the SEM treatment. After washing 3 times with PBS, overnight fixation with 2.5% glutaraldehyde was performed, and secondary fixation with 1% osmium tetroxide for 2 h. The fixed bacterial cells were washed thrice, followed by dehydration using a graded acetone series of 10%, 30%, 50%, 70%, 90% and 100% for 10 min each. After being placed in absolute acetone 2 times for 10 min each, infiltration of samples by 1:1, 1:2 and 2:1 mixture of acetone and epoxy resin for 3 h each was performed. Thereafter, the samples were transferred to pure epoxy resin 3 times for 3 h each time. Samples were embedded using a flat embedding mold. The castings were polymerized at 70 °C for 8 h. Then, ultrathin sections at a thickness of 70 nm were obtained using an ultramicrotome (Leica EM UC7; Leica). The samples were poststained with 5% uranyl acetate and lead citrate. Specimens were examined with a transmission electron microscope (Hitachi HT7700).

Induction of resistance by serial passages at MIC. Serial 24-h passages of MIC were performed in
96-well plates in order to evaluate the development of resistance after long exposure to D-AP19 or meropenem, as previously described with some modification 29 . A. baumannii ATCC 19,606 was adjusted to approximately 10 6 CFU/ml in MHB. The bacteria were treated with D-AP19 or meropenem at concentrations of 0.98 to 250 µg/ ml, and 0.23 to 125 µg/ml, respectively. After 24 h incubation, bacterial suspensions of the wells at half-MIC concentration of tested agent were taken and washed 2 times with MHB to prepare the next initial-passage of bacteria. Twenty repeat passages were performed for each tested compound. Along with the test, a negative control was included: the MIC value of bacteria cultured in MHB-deionized water after 20 passages.
Statistical analysis. The data from three independent experiments were exhibited as mean ± standard deviation (SD); differences with 95% confidence levels (p < 0.05) were considered statistically significant. Oneway ANOVA with Tukey's Post Hoc Test analyzed via GraphPad PRISM software (version 7.0, GraphPad Software, California, USA) were applied to evaluate differences between control and tested groups.

Results
Peptide design and characteristics. The parent peptide, A3P7, was designed by flipping of the hybrid analogue peptide, P7A3, which was successfully modified from the sequence of cathelicidin and aurein 18 . A3P7 showed very high toxicity toward human red blood cells ( Fig. 1) and therefore was serially truncated (from the C-terminal end) to obtain a peptide retaining potent antimicrobial potency but with reduced toxicity. As shown in Table 1, eleven derivatives were obtained and their physicochemical characteristics were in silico characterized. All truncated peptides were chemically synthesized for in vitro antimicrobial testing. Among the truncated derivatives, the 19-amino acid peptide, AP19, showed equal hydrophobicity to that of the parent peptide, A3P7, with a high positive net charge (+ 9). ESI-MS was used to confirm the molecular weights of peptides ( Table 1). The theoretical molecular weight (MW) of each peptide was consistent with its measured MW. This suggested that peptide synthesis was successful.
Candidate peptide AP-19 exhibited lowest GM value and highest therapeutic index. As shown in Table 2   www.nature.com/scientificreports/ (Table 2), AP19 had the greatest antimicrobial activity against both Gram-negative and -positive bacteria (had the lowest GM values). The hemolytic activity of peptides against fresh hRBCs was determined at various concentrations (0.98 to 250 µg/ml) to indicate their toxicity on mammalian cells (Fig. 1). Melittin, the positive control, completely lysed hRBCs at a concentration of 3.91 µg/ml. The parent peptide (A3P7) had obvious hemolytic effects displayed in a concentration-dependent manner. Of truncated peptides, AP12 to AP19 showed a marked decrease in hemolytic effects on hRBCs at all concentrations, with MHC (minimum concentration that causes 10% hemolysis) greater than 250 µg/mL. These results highlighted an advantage of truncation, a significantly reduced toxicity of the peptides. The therapeutic index (TI) is an important parameter to identify a candidate peptide, reflecting a balance between efficacy and safety 30 . It was calculated as the MHC value divided by GM value (Table 2) 23 . For Gram-negative bacteria, AP18 and AP19 exhibited the highest TI values of the designed peptides. For Grampositive bacteria, AP19 showed the highest TI value. The TI value of AP19 against Gram-negative and -positive bacteria was 332 and 294 times higher, respectively, than that of A3P7 parent peptide. D-AP19 displayed potent antibacterial activity toward MDR and XDR A. baumannii along with improved stability. AMPs have failed in clinical trials due to their low stability when exposed to proteolytic enzymes and human plasma. Due to chirality mismatch with active sites of enzymes, D-enantiomers are usually resistant to proteolysis by endogenous enzymes, especially proteases 31 . Thus, to increase its stability toward proteolytic enzymes peptide AP19 was further modified by D-amino acid substitution and designated as D-AP19. It retained the same MIC and MBC values against A. baumannii (a first priority pathogen on the WHO list) in contrast to AP19 ( Table 3). The MIC and MBC values of AP19 against A. baumannii ATCC 19,606 increased more than 32-fold (from 7.81 to more than 250 µg/mL) in the presence of trypsin and proteinase K, indicating a complete loss of AP19 anti-microbial activity (Table 4). D-AP19 displayed tolerance to trypsin and proteinase K as its MIC was not changed (7.81 µg/mL) in the presence of these two proteases when compared to untreated peptide. With pepsin treatment, the MICs of AP19 and D-AP19 were fourfold (from 7.81 to 31.25 µg/mL) and twofold (from 7.81 to 15.63 µg/mL) increased, respectively. These finding suggested that D-AP19 exhibited more stability to proteolytic enzymes than did AP19. Table 2. Antibacterial activity, minimum hemolytic concentration (MHC), geometric mean of MIC value (GM) and therapeutic index (TI) of hybrid peptide derivatives. a MHC, minimum hemolytic concentration, indicated the peptide concentration that caused 10% hemolysis of human red blood cells (hRBC). When there was less than 10% hemolysis detected at concentration of 250 µg/ml, a value of 500 µg/ml was applied to calculate the therapeutic index. When there was less than 10% hemolysis detected at concentration less than 0.98 µg/ml, a value of 0.49 µg/ml was applied to calculate the therapeutic index. b GM (Gr.-strains) indicated the geometric mean of MIC value from all Gram-negative bacterial strains. When MIC was not detected at 250 µg/ml, a value of 500 µg/ml was applied to calculate the therapeutic index. c Therapeutic index (Gr.strains) is the ratio of the MHC to the geometric mean of MICs from Gram-negative bacterial strains. d GM (Gr. + strains) indicates the geometric mean of MIC values from all Gram-positive bacterial strains. When MIC was not detected at 250 µg/ml, a value of 500 µg/ml was applied to calculate the therapeutic index. e Therapeutic index (Gr. + strains) is the ratio of the MHC to the geometric mean of MICs from all Gram-negative bacterial strains. www.nature.com/scientificreports/ The nosocomial pathogen A. baumannii is responsible for various types of infections in the healthcare setting, including pneumonia, bacteremia and meningitis 32 . Its threat is growing as indicated by steady increase in prevalence of MDR or XDR A. baumannii over the past decade 32 . In this study, D-AP19 showed high antibacterial activity against all MDR and XDR clinical isolates of A. baumannii with MICs and MBCs ranging from 3.91 to 15.63 µg/mL (Table 3). For comparison, the antibacterial activities of meropenem, levofloxacin, minocycline and colistin, the commonly used antibiotics for in-hospital A. baumannii infections, were also examined against these drug-resistant bacteria. The antibacterial activity of AP19 and D-AP19 against meropenem-resistant A. baumannii (MIC value of 3.91 to 15.63 µg/mL), was 16 to 32 times higher than that of meropenem (MIC value of 62.5 to 125 µg/mL).
Some components of human plasma, such as binding or blocking proteins and peptidases, can inactivate peptides or degrade their amino acid sequences 33 . After incubation with human plasma, the MIC and MBC values of AP19 were 32-fold increased from 7.81 to 250 µg/mL, whereas the MICs of D-AP19 were only twofold increased (from 7.81 to 15.63 µg/mL) ( Table 4). These results indicated that D-AP19 was more stable in human plasma than was AP19.

D-AP19 displayed low toxicity to mouse fibroblast cells and low hemolytic activity to human
RBCs. The cytotoxic activity of AP19 and D-AP19 was investigated by measuring viability of L929 mouse fibroblast cells after exposure to peptide. The membranolytic melittin was included in this test as a positive control. Melittin had robust cytotoxicity in a concentration-dependent manner, indicated by the decrease of cell viability to less than 5% at a peptide concentration of 62.5 µg/mL (Fig. 2). In contrast, AP19 showed no toxicity as indicated by 100% cell viability for all tested peptide concentrations. At 1 × and 2 × MIC of D-AP19, 100% cell viability was observed. In contrast, the cell viability of D-AP19 at peptide concentrations of 62.5, 125 and 250 µg/mL was significantly greater than that of the negative control (p > 0.05). But based on the ISO standards 34 , D-AP19 (at concentrations of 62.5 and 125 µg/mL) was not considered a cytotoxic agent since cell viability was more than 70%. The hemolytic activity of D-AP19 was decreased compared with AP19. Hemolysis was 7.25% and 3.28% at a concentration of 250 µg/mL for AP19 and D-AP19, respectively. At the MIC of D-AP19 against A. baumannii ATCC 19,606, hemolysis was only 0.58%.  15.63 µg/mL. These results suggested that both L-and D-form of AP19 peptide exhibited antibacterial activity despite the presence of most physiologic salts. But of note, NaCl reduced the antibacterial activity of both peptides.

D-AP19 and its L-enantiomers formed an α-helical amphipathic conformation in membrane mimetic environments.
The secondary structure of peptides in PBS as well as in the membrane mimetic environments (30 mM SDS micelles in PBS and 50% TFE in PBS) was determined by CD spectroscopy. The CD spectra analyses were interpreted in accordance with published data 35,36 . As shown in Fig. 3, the spectra of all peptide derivatives (A3P7, AP13, AP16, AP17, AP18, AP19 and D-AP19) in PBS showed obvious negative signals at around 200 nm [characteristic of a random coil (or unordered) structure], while melittin showed the weak signal of an α-helical structure. In 30 mM SDS and 50% TFE, melittin and A3P7 formed α-helical conformations with maximum signals at 192 nm and two minimum signals around 208 and 222 nm. The truncated peptides (AP13, AP16, AP17, AP18 and AP19) formed α-helical conformations in 30 mM SDS micelles. In 50% TFE, peptides AP16, AP17, AP18 and AP19 adopted α-helical configurations, while AP13 formed a weak α-helical structure (Fig. 3). D-AP19 displayed characteristic signals for a random coil structure in PBS, and strong α-helical configurations in 30 mM SDS micelles and 50% TFE. The CD spectra of D-enantiomers of AP19 were mirror images of those of its L-enantiomer. These results indicated that A3P7, AP19, AP18 and AP17, as well as D-AP19, transformed from an unordered random coil structure in aqueous environment to an α-helical amphipathic conformation in membrane mimetic environments.

D-AP19 exhibited rapid killing of A. baumannii. The time course for bactericidal activity of D-AP19
against A. baumannii ATCC 19,606 was determined (shown in Fig. 4). Compared to growth of the bacterial negative control, D-AP19 decreased cell viability of A. baumannii more than 10 2 , 10 3 and 10 4 CFU/mL within 0.25, 0.5 and 1 h, respectively. Within 4 h, D-AP19 completely killed all A. baumannii and no regrowth was observed after 24 h of observation. As seen in Table 3, the MIC of D-AP19 against each of the standard and drug-resistant strains of A. baumannii was equal to its MBC. Therefore, complete killing of the bacterial inoculum after 4 h Figure 2. The cytotoxicity of AP19 and D-AP19 against L929 mouse fibroblast cells compared with that of melittin (a positive control). Three independent experiments were performed and the data are presented as mean ± SD. The statistical analyses utilized one-way ANOVA and Tukey's test at p-value < 0.05 (GraphPad Prism7). Star indicates a significant difference from negative control (p-value < 0.05).

Ultrastructural evidence of D-AP19's disruption and distortion of bacterial cell membranes.
Morphologic changes of A. baumannii ATCC 19,606 was directly observed using SEM. On untreated bacterial cells the membrane surfaces were smooth and intact ( Fig. 6A and B). On the contrary, membranes of bacterial cells treated with 0.5 × MIC of D-AP19 for 15 min were significantly damaged. Rough surfaces and shrinkage of bacterial cells were observed (Fig. 6E,F). Multi-blebbing of many cells with pili-like structures were obvious (Fig. 6C,D), as well as distortion and disruption of bacterial cell membranes (Fig. 6E,F). The effects of D-AP19 on the morphology and intracellular contents of A. baumannii ATCC 19,606 were also assessed by using TEM. A. baumannii grown in PBS without D-AP19 (negative control) displayed intact cell membranes and clearly visible cell walls (Fig. 7A and B). After 15 m incubation with D-AP19 (0.5 × MIC), ultrastructural alterations were observed, including obvious clumped material attached to inner membranes, apparent cytoplasmic clear zones, blurred cell membranes and intracellular leakage (Fig. 7C-F).

D-AP19 did not induce resistance in A. baumannii in vitro.
The high propensity of antibiotic therapy to induce bacterial resistance is a major obstacle to continued efficacy of antibiotics, harming healthcare D-AP19 (0.5 × MIC) or meropenem was evaluated through 20 serial passages by broth microdilution assay (Fig. 8). Bacteria serially grown in drug-free medium for 20 passages were used as negative controls. The MICs of D-AP19 and meropenem against these negative control bacteria were equal to the MICs at the first passage, suggesting that A. baumannii ATCC 19,606 did not develop resistance in these non-drug conditions. At day 4, 6 and 11, MICs of meropenem were 2-, 4-and eightfold increased, respectively. After passage 20, meropenem exhibited a 16-fold increase in MIC compared to the first passage, resulting in a final MIC of 15.63 µg/mL (a 15-fold increase). Remarkably, D-AP19 antimicrobial activity did not decrease over the 20 passages, retaining an MIC value of 7.81 µg/mL.

Discussion
Nosocomial bacterial infection is an important cause of morbidity and mortality among patients with serious health problems 2,37 . According to a WHO announcement, carbapenem-resistant Acinetobacter baumannii is classified as a priority pathogen which urgently needs new and effective therapeutic agents 3 . Antimicrobial peptides, cationic peptides with innate immune properties, possess multi-action mechanisms against pathogenic bacteria, including membrane attack, intracellular alteration and/or immunomodulation 9 . They have a low propensity to induce bacterial resistance and can kill drug-resistant bacteria 8 . Although these advantages are well-known, AMPs also have some weaknesses. For instance, they may be toxic to human cells and lose activity in the presence of proteolytic enzymes, salts, serum and/or plasma 9 . But of note, amino acid sequence modifications can overcome these shortcomings and generate potentially successful peptide-based drugs. Thus, the design and modification of membrane-penetrating AMPs is a prominent strategy to obtain novel antimicrobial agents with satisfactory efficacy and toxicity features. www.nature.com/scientificreports/ We previously reported a hybrid peptide, a modified α-helix cathelicidin (P7) and aurein (A3) designated as P7A3, with promising antimicrobial activity and potential applications 18 . In the present study, a novel hybrid peptide (A3P7) was obtained using a flipping technique or reverse conjugation, in the hybridizing of A3 with P7 at the C-terminus. This peptide had imperfect amphipathicity, high positively charge and hydrophobicity, and exhibited broad-spectrum antibacterial activity against both Gram-negative and -positive bacteria comparable to that of P7A3, along with moderate hemolytic activity (compared to that of melittin). The active site of α-helical AMPs is mostly located at the N-terminus which normally interacts with and penetrates through the bacterial membrane 38 . To reduce sequence length and toxicity against human red blood cells, A3P7 was sequentially truncated at the C-terminus. Among the resulting truncated peptides, AP19 showed improved  www.nature.com/scientificreports/ more difficult 39 . Moreover, they are usually more hemolytic than their shorter derivatives 9 . Multiple physiologic characteristics of AMPs, including number of amino acid residues (size), net charge, hydrophobicity, amphipathicity and helicity, are responsible for their antimicrobial and hemolytic activities 9,40 . But modification of AMPs has complicated effects, since changing one factor will affect others, and so multiple activities must be monitored as changes are made 41 .
The therapeutic index is a useful parameter in identifying candidate peptides, as it reflects a balance between the efficacy and safety of candidate agents 30 . Higher TI values indicate a peptide's favorable safety and promising efficacy. Compared to A3P7, TI values of all truncated peptides were increased as a result of reduced hemolytic activity. The truncated peptide AP19 exhibited the highest TI value against both Gram-negative and Grampositive bacteria with the values of 69.8 and 85.32, respectively. This resulted from strong antibacterial activity and weak hemolytic activity when compared to that of the parent peptide A3P7, reflecting greater therapeutic benefit without increased toxicity. These results suggested that C-terminal truncation improved the selectivity of peptides toward bacterial cells more than hRBCs. The successful development of AP19, as indicated by its high antibacterial activity and low hemolytic activity, opened interesting opportunities for AMP development and practical application.
An important obstacle responsible for past failures of AMPs in clinical trials is their loss of antibacterial activity in the presence of proteolytic enzymes and human plasma 42 . AP19 showed low stability in pepsin, trypsin, proteinase K and human plasma. Due to chirality mismatches with active sites of enzymes, D-enantiomers are usually resistant to proteolysis by endogenous enzymes, especially proteases 31 . To improve the stability of our peptide, AP19 was further modified by D-amino acid substitution (designated as D-AP19). The antimicrobial activity of D-AP19 against A. baumannii ATCC 19,606 was similar to that of AP19, suggesting D-amino acid incorporation did not affect this activity of the peptide, a finding similar to that of other studies 43,44 . D-AP19 retained potent antibacterial activity after exposure to all tested commercial proteases (pepsin, trypsin and proteinase K) and human plasma with MICs from 7.81 to 31.25 µg/mL. This was consistent with other research that D-amino acid substitution is an effective strategy to improve peptide stability against proteases and human plasma 43,44 .
WHO has announced an urgent need for development of new antimicrobial agents to fight against MDR bacteria, especially multidrug-resistant A. baumannii 3 . AP19 and D-AP19 exhibited the same level of antibacterial activity against both reference and resistant strains of MDR and XDR A. baumannii, whereas traditional antibiotics (meropenem and levofloxacin) required higher concentrations to kill these bacteria. Thus, our designed peptide was effective against resistant strains of clinically isolated A. baumannii and its antibacterial activity was not significantly affected by the bacterial resistance phenotype [43][44][45] .
Others have reported that physiologic concentrations of serum salts diminish the antimicrobial activity of several peptides, and that this is due to the interruption of the electrostatic interactions between positively charged AMPs and negatively charged microbial membranes in the presence of salt ions 46 . AP19 retained antibacterial activity in the presence of all tested serum ions, except Na + . Peptide D-AP19 retained its antibacterial activity in the presence of Mg 2+ and NH 4 + , while its activity was slightly weakened by K + , Zn + , Ca 2+ and Fe 3+ . In the presence of Na + , the antibacterial activity of AP19 and D-AP19 was greatly decreased. The results revealed that D-amino acids substitution of AP19 could not enhance the peptide stability in the presence of serum salts. RR4 and its D-enantiomer showed similar antimicrobial activities in the presence of Na + , Ca 2+ and Mg 2+ ions 43 .
The toxicity of D-AP19 against hRBCs and L929 mouse fibroblast cells was compared to that of AP19. D-AP19 showed slightly decreased hemolytic activity, thus a better safety profile than AP19. Similarly, the D-forms of Bac and GL13K peptides have less hemolytic activity than their L-forms 45,47 . On the contrary, D-AP19 at high concentrations showed greater toxicity on L929 cells than did its L-form. Similarly, the D-enantiomers (Pro9-3D and Pro10-1D) of peptide Pro9-3 and Pro10-1 showed greater toxicity against mouse macrophage RAW264.7 cells than did their L-enantiomers 48 . These results suggested that D-AP19 was not toxic against human red blood cells but was against mouse cells. www.nature.com/scientificreports/ The helix forming ability of peptides in membrane mimetic environments was correlated with their antimicrobial activity 49,50 . Melittin, A3P7, AP19, AP18, AP17, AP16 and D-AP19 formed α-helical amphipathic structures in both 30 mM SDS and 50% TFE with high and moderate antimicrobial activity. AP13 failed to form a helical conformation in 50% TFE, but did so in 30 mM SDS; these had low antimicrobial activity. From previous studies, C-terminal truncated derivatives of P5 peptide (P5-CT2) lose the ability to form α-helices in 30 mM SDS, but retain this in 50% TFE; antimicrobial activity is lower than that of P5 25 . In parallel, the helical content of peptides when interacting with 50% TFE was correlated with toxicity against hRBCs 49 . A3P7 possessed 58.4% α-helical content and showed an MHC value of 1.95 µg/mL. Previous studies revealed that the hemolytic activity of peptides generally decreased with reduction of peptide helicity 22,49 .
AMPs have multiple mechanisms of action aimed at multiple targets in bacteria, including cytoplasmic membranes and intracellular targets 51 . Most AMPs interact with and disrupt bacterial membranes, causing membrane depolarization and permeabilization, and leading to cell dead 10,52 . However, the exact mechanism of these actions is unclear 52 . Membrane depolarization and permeabilization after treatment by D-AP19 were analyzed using BOX and PI fluorescent dyes, respectively. The results showed that these effects were not time dependent. Within 15 min, D-AP19 had induced bacterial membrane depolarization and permeability. This directly correlated with D-AP19's fast killing at MIC, with approximately 30-40% of A. baumannii killed within 15 min. The SEM and TEM studies showed that D-AP19 damaged A. baumannii via membrane-based mechanisms and leakage of intracellular contents, with additional unknown mechanisms resulting in the formation of inclusion bodies which accumulated at the inner bacterial membranes. The membrane depolarization and permeabilization is one of killing mechanism of D-AP19. Many AMPs have been reported to depolarize and permeabilize the membranes of A. baumannii, resulting in intracellular leakage 44,53 . The limitation of this study was that the mechanism(s) of action of D-AP19 toward intracellular target in A. baumannii ATCC 19,606 was not clearly identified.
In the clinical use of antibiotics to treat nosocomial infection, development of bacterial resistance interrupts the efficiency of treatment. The serial incubation of bacteria in half-MIC of peptide is a widely used method of inducing resistance. Using the method, D-AP19 retained its antibacterial activity at 7.81 µg/mL without increase through 20 culture passages, while under the same conditions meropenem showed an increase of MIC. These results are similar to those with AA139 and ZY4 peptides 29,54 . Induction of resistance to antibiotics is often associated with a single mechanism of action 54 . D-AP19, with its multi-mechanisms of action, showed a very low propensity to induce resistance, and so might be an effective way to combat the development of resistant bacteria 54 .

Conclusion
This study emphasized the potential to develop a protease tolerant, membrane-active peptide (D-AP19), as a therapeutic agent especially against multi-or extensively-drug resistant A. baumannii with no toxicity to human cells. In addition, the antibacterial activity of D-AP19 was not affected by the resistance phenotype of the target bacteria. Our candidate peptide (D-AP19) rapidly killed bacteria through interactions with membrane targets and forming α-helical structures. Subsequently, membranes became depolarized and permeable, leading to leakage of intracellular contents and membrane disruption, and ultimately cell death. Importantly, D-AP19 also displayed a low tendency to induce bacterial resistance. This study exploited the advantages of peptide modification for improvement of AMPs in terms of antibacterial activity, safety and stability. With further study of efficacy and safety carried out in vivo, peptide D-AP19 might be shown to have potential for human therapy, and join the challenging fight against MDR or XDR A. baumannii.

Data availability
All of the data produced throughout the research are contained in this article and the supplementary information file.