Bovine NK-lysin peptides exert potent antimicrobial activity against multidrug-resistant Salmonella outbreak isolates

Multidrug-resistant (MDR) Salmonella is a threat to public health. Non-antibiotic therapies could serve as important countermeasures to control MDR Salmonella outbreaks. In this study, antimicrobial activity of cationic α-helical bovine NK-lysin-derived antimicrobial peptides was evaluated against MDR Salmonella outbreak isolates. NK2A and NK2B strongly inhibited MDR Salmonella growth while NK1 and NK2C showed minimum-to-no growth inhibition. Scrambled-NK2A, which is devoid of α-helicity but has the same net positive charge as NK2A, also failed to inhibit bacterial growth. Incubation of negatively charged MDR Salmonella with NK2A showed increased Zeta potential, indicating bacterial-peptide electrostatic attraction. Confocal and transmission electron microscopy studies revealed NK2A-mediated damage to MDR Salmonella membranes. LPS inhibited NK2A-mediated growth suppression in a dose-dependent response, suggesting irreversible NK2A-LPS binding. LPS-NK2A binding and bacterial membrane disruption was also confirmed via electron microscopy using gold nanoparticle-NK2A conjugates. Finally, NK2A-loaded polyanhydride nanoparticles showed sustained peptide delivery and anti-bacterial activity. Together, these findings indicate that NK2A α-helicity and positive charge are prerequisites for antimicrobial activity and that MDR Salmonella killing is mediated by direct interaction of NK2A with LPS and the inner membrane, leading to bacterial membrane permeabilization. With further optimization using nano-carriers, NK2A has the potential to become a potent anti-MDR Salmonella agent.


Results
Antimicrobial activity of NK-lysin peptides against MDR and non-MDR Salmonella isolates. The antimicrobial effectiveness of bovine NK-lysin-derived peptides was assessed against MDR Salmonella I 4, [5],12:i:-pork outbreak isolate USDA15WA-1 (SX238) and a non-MDR Salmonella isolate (SB377) which is known as universal killer (UK1) ( Table 1) in LB broth medium. Four synthetic 30-mer peptides (i.e., NK1, NK2A, NK2B, and NK2C) corresponding to the functional region of bovine NK-lysin (helix-2-loophelix-3) were tested in growth curve assays. As shown in Fig. 1a and Supplementary Fig. S1, NK1 did not show Originally isolated from the spleen of a chick orally inoculated 3 days earlier with a highly virulent S. typhimurium isolate retrieved from an infected horse in 1991 Our isolate was passaged through a pig and isolated from the ileocecal lymph node 60 www.nature.com/scientificreports/ any Salmonella SX238 and SB377 growth inhibition activity at the tested peptide concentrations (highest up to 20 µM, Fig. 1a). Although 10 µM and 20 µM NK2C initially inhibited Salmonella SX238 and SB377 growth, bacteria recovered and grew at approximately 7 h and 20 h, respectively (Fig. 1a). In contrast, both NK2A and NK2B were highly effective and completely inhibited growth of Salmonella SX238 and SB377 at ≥ 5 µM concentration (Fig. 1a). Since NK2A peptide had a higher antimicrobial activity than NK2B (growth inhibition at 2 µM NK2A concentration for up to 8 h), this bovine NK-lysin peptide was selected for further studies, and the SX238 isolate was chosen for further studies with NK2A. Bactericidal activity of NK2A during 3 h incubation was also tested in PBS and LB broth, and all three NK2A concentrations (2, 5, 20 µM) were highly effective against Salmonella SX238, with greater than 97% and 99% killing, respectively (P = 1.000) (Fig. 1b).
CD spectroscopy of Sc-NK2A. In this study, the secondary structure of scrambled NK2A (Sc-NK2A) in a bacterial membrane-mimicking environment was determined using CD spectroscopy. Previous CD spectral , and optical density readings at 600 nm (OD 600nm ) were recorded using an automated growth curve reader. One representative growth graph out of at least three independent experiments is shown. (b) Different concentrations of NK2A were incubated with Salmonella isolate SX238 at 37 °C for 3 h in PBS (filled bars) or in LB broth (open bars), serially diluted and plated on XLT plates. Following overnight incubation, bacterial colonies were enumerated. Data represent mean percentages of (bacterial) killing ± SD from three independent experiments. P = 1.0000 as determined by ANOVA test using GLM procedure from SAS software, Version 9.4 of the SAS System for Windows. www.nature.com/scientificreports/ profile analyses confirmed all four bovine NK-lysin peptides to have α-helical conformation when bound with liposomes 24 . The amino acid sequence in NK2A was scrambled and chemically synthesized as a control peptide (Sc-NK2A). Sc-NK2A is predicted to have no α-helical structure but with the same net positive charge as original NK2A (net charge = + 7.9). Both peptides in phosphate buffer (without liposomes) showed CD spectra indicative of random coil structures (Fig. 2a). NK2A peptide in liposomes showed a CD profile characteristic of predominantly α-helical structure (maximum near ~ 192 nm with double minima near ~ 208 nm and ~ 222 nm) (Fig. 2a). In contrast, Sc-NK2A did not show a CD profile indicative of α-helical structure (Fig. 2a), suggesting that the rearrangement of amino acids resulted in elimination of secondary α-helical structure.
Cytotoxicity studies with NK-lysin peptides. To assess the toxicity of NK-lysin peptides, a hemolytic assay and a mouse study were performed. It has previously been reported that all four 30-mer NK-lysin-derived peptides have minimal cytotoxic effect against cattle RBCs 24 . To determine whether Sc-NK2A has hemolytic activity, a hemoglobin release assay was performed with cattle RBCs. Both NK2A and Sc-NK2A at the tested concentrations (2-20 µM) showed < 2% hemolytic activity with no significant differences observed between the two peptides (P = 0.1650) (Fig. 2b). Mice intraperitoneally injected with 100 µg of NK-lysin peptides (NK1, NK2A, NK2B, NK2C, or Sc-NK2A) did not show signs of toxicity or death during the 2-week observation period. These findings suggest that NK-lysin peptides have minimal toxic effects to both bovine red blood cells and mice at the tested concentrations.
Antimicrobial activities of Sc-NK2A peptide. Bacterial growth inhibition assays were performed with Salmonella isolate SX238 using both NK2A and Sc-NK2A peptides (Fig. 2c). NK2A completely inhibited growth of SX238 at 10 5-7 CFU/mL (or 10 4-6 CFU/well) during a 24 h growth period, and partial growth inhibition was observed (up to 9 h) at 10 8 CFU/mL (or 10 7 CFU/well). Sc-NK2A failed to inhibit bacterial growth even at the lowest bacterial concentration (10 4 CFU/well). Using additional MDR Salmonella outbreak isolates originating from cattle (SX231) and turkey (SB395) along with SX238 and SB377 at 10 5 CFU/well, the antimicrobial activity of NK2A and Sc-NK2A was compared. NK2A inhibited the growth of the four Salmonella isolates at or above the 5 µM concentration, although three of the four isolates were sensitive to a final NK2A concentration of 3.5 µM (Fig. 3). However, even 20 µM Sc-NK2A failed to show Salmonella growth inhibition activity (Fig. 3).
Effect of NK2A on Zeta potentials of MDR Salmonella. Bacterial Zeta potential has recently been shown to serve as a proxy for membrane permeabilization 35 . Accordingly, Zeta potential measurements of NKlysin peptides and Salmonella were evaluated. Cationic NK-lysin peptides including Sc-NK2A showed positive Zeta potential values in the range of 16.3-24.8 mV (Table 2). In contrast, three MDR and one non-MDR Salmonella isolates had negative Zeta potential values, which is attributed to the anionic LPS-outer membrane (Table 3). Bacterial Zeta potential changes following incubation with NK2A (2 µM for 20 min) were assessed. Alteration of Salmonella Zeta potentials was observed following incubation with NK2A; the negative Zeta potentials of all four Salmonella isolates changed to positive Zeta potential values (ranging from 6.1-16.0 mV) suggesting NK2A-LPS (Salmonella) and/or NK2A-cytoplasmic membrane binding (Table 3). MDR Salmonella isolate SB395 with 2 µM Sc-NK2A increased Zeta potential value (7.8 mV), indicating that Sc-NK2A also bound to Salmonella membranes (Table 3). However, as shown in Fig. 3, Sc-NK2A at 20 µM did not inhibit Salmonella growth, suggesting that the scrambled, non-pore forming peptide can associate closely with the cell surface to change the observed bacterial surface charge to positive, but the presence of the α-helical structure is critical for anti-bacterial activity.
Effects of NK2A on MDR Salmonella morphology. To morphologically evaluate the dead MDR Salmonella SX238 following incubation with NK2A, confocal laser scanning microscopy (CLSM) and transmission electron microscopy (TEM) studies were performed. The LIVE/DEAD (BacLight) bacterial viability staining method was used to differentiate live (stained with SYTO 9, green) from dead (stained with propidium iodide (PI), red) bacteria. Primarily SYTO 9-stained bacteria (green) and only a few PI-stained bacteria (red) were observed in both control (PBS; Fig. 4a) and Sc-NK2A (Fig. 4b) treated samples, indicating that Sc-NK2A did not have detectable antimicrobial activity. In contrast, many dead (red) Salmonella SX238 were detected in NK2A treated samples, further confirming the growth assay findings that NK2A is bactericidal to Salmonella (Fig. 4c).
To visualize ultrastructural changes to Salmonella SX238 outer-(LPS) and inner-(cytoplasmic) membranes following incubation with NK2A, bacterial samples were evaluated using TEM. Salmonella SX238 in PBS (control, Fig. 4d) or with Sc-NK2A (Fig. 4e) were normal bacilli-shaped bacteria with intact outer-and inner-membranes and typical electron-dense cytoplasm. The periplasmic spaces in both samples were thin and uniform. These observations suggest that the bacteria were alive, and Sc-NK2A did not cause detectable damage to the bacterial membranes. In contrast, SX238 treated with NK2A had extensively damaged outer-and inner-membranes and cytosolic content leakage (Fig. 4f,g). Due to the loss of membrane integrity and cytosolic contents, the bacteria were smaller in size, lacked the typical bacilli morphology and appeared collapsed. NK2A-treated SX238 had slightly wavy inner membranes (Fig. 4f,g), wider periplasmic spaces (Fig. 4f,g and 6d) and were visibly nonuniform. Collectively, these findings suggest that NK2A disrupted both the outer-and inner-membranes of Salmonella. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 3. Antimicrobial activity of NK2A and Sc-NK2A against Salmonella isolates. Three MDR Salmonella outbreak isolates (SX238, SX231 and SB395) and one non-MDR Salmonella isolate (SB377) (10 6 CFU/mL) were incubated with different concentrations of NK2A, Sc-NK2A or Luria-Bertani broth (LB) and optical densities (OD 600nm ) were recorded using an automated growth curve reader over a 24 h period. One representative graph (for each peptide and each isolate) out of at least three independent experiments is shown. www.nature.com/scientificreports/ cytoplasmic membranes are compromised 20 . With similar features as NK2A, cyclic polycationic polymyxin B was selected as a positive control for the assay because polymyxin B is a peptide and a rapid-acting bactericidal antibiotic which interacts with bacterial LPS as well as the cytoplasmic membrane 36 . Growth curve assays with polymyxin B (2-512 µg/mL) showed that as little as 2 µg/mL was sufficient to completely inhibit Salmonella SX238 growth during the 24 h incubation (Fig. 5a). No PI uptake signal was observed with Salmonella SX238 incubated with polymyxin B at concentrations even as high as 100 µg/mL during a 30 min observation period ( Fig. 5b) as well as 6 h observation period ( Supplementary Fig. S2). Furthermore, Salmonella incubated with Sc-NK2A or PBS (negative control) did not show PI uptake signals (Fig. 5b), indicating that neither polymyxin B nor the scrambled NK2A peptide disrupted the cytosolic membranes. In contrast, rapid PI uptake was observed for Salmonella SX238 incubated with NK2A (5 µM), and PI fluorescent signal intensity steadily increased during the 30 min incubation period, suggesting that the bacterial cytoplasmic membrane was disrupted (or permeabilized) by NK2A (Fig. 5b).

Characterization of nanoparticles.
To test the compatibility of NK2A with polymeric nano-carrier delivery devices, a small library of polyanhydride nanoparticle formulations was synthesized. Empty 20:80 CPTEG:CPH, 10:90 CPTEG:SA, and 20:80 CPH:SA nanoparticle formulations had mean diameters of ~ 239-341 nm (Table 2). NK2A-loaded formulations had similar size distributions to the corresponding empty formulations (Table 2). Morphologically, the CPTEG:SA and CPH:SA nanoparticles tended to be spherical and less aggregated, while the CPTEG:CPH nanoparticles tended to be less spherical and slightly more aggregated when visualized by SEM (Fig. 7a).  www.nature.com/scientificreports/ Payload type dramatically affected nanoparticle release kinetics in PBS, pH 7.4. Kanamycin-loaded nanoparticles showed a burst-dominated release profile, with approximately 90% release within 1 day and little drug released over the following 6 days (Fig. 7b). This result could indicate poor compatibility of the relatively www.nature.com/scientificreports/ hydrophilic drug within the hydrophobic polyanhydride polymer matrix, resulting in kanamycin localization at the particle surface. By contrast, NK2A formulations showed a slow, sustained release, releasing < 35% of their payload over 1 week (Fig. 7b). This slow release is likely a result of the hydrophobicity of the NK2A peptide enabling better mixing with the polymer within the nanoparticles. This correlates with a more homogeneous distribution of NK2A within the particles, allowing polymer degradation and surface erosion to control release. All NK2A-and kanamycin-loaded nanoparticles showed moderate encapsulation efficiencies between 34.3 and 47.3%. Each empty nanoparticle chemistry showed negative Zeta potential values, attributable to dicarboxylic acid residues that are produced at the surface after hydrolysis of the anhydride bond 37 (Table 2). However, NK2A encapsulation produced a profound, sign-inverting increase in observed Zeta potentials, likely due to the cationic nature of the NK2A peptide (Table 2).
Antimicrobial activity of NK2A nanoparticles. To assess the antimicrobial activity of NK2A nanoparticles, MDR Salmonella SX238 (10 6 CFU/ml) was incubated (3 h) with NK2A nanoparticles (5 µM final concentration). Efficient Salmonella SX238 killing was observed with 20:80 CPTEG:CPH kan nanoparticles (final kanamycin concentration = 16 µg/mL) (Fig. 7c). Both 10:90 CPTEG:SA and 20:80 CPH:SA NK2A-loaded nanoparticles showed 100% Salmonella SX238 killing, while 20:80 CPTEG:CPH NK2A-loaded nanoparticles exhibited ~ 80% Figure 5. Assessment of propidium iodide (PI) uptake by Salmonella incubated with NK2A, Sc-NK2A and polymyxin B. (a) Growth curve graphs (OD 600nm ) were generated using an automated growth curve reader for Salmonella outbreak isolate SX238 after incubation with different concentrations of polymyxin B (0-512 μg/ mL). One representative graph out of three independent experiments is shown. (b) Salmonella (SX238) was preincubated with PI followed by the addition of NK2A, Sc-NK2A (5 μM final concentration) or polymyxin B (100 μg/mL concentration), and PI signal was measured by a fluorescent microplate reader using flex mode with 15 s intervals for up to 30 min. PI uptake signals are shown as relative fluorescent units (RFUs). Bacteria incubated with PBS was used as a negative control. One representative PI uptake graph out of three independent experiments is shown. www.nature.com/scientificreports/ SX238 killing (Fig. 7c) (P < 0.0001). This differential Salmonella SX238 killing could indicate polymer chemistry interactions with the NK2A peptide, affecting its capacity to localize at the bacterial surface, or alternatively polymer-bacteria interactions. These considerations should be explored further in future research.

Discussion
Salmonella is a major causative agent of bacterial foodborne illness in the U.S. 3,38 . In a 2019 United States antibiotic resistance threats report, the CDC classified both drug resistant nontyphoidal and typhoidal Salmonella as serious threats to global health 2 . Approximately 10% of Salmonella are multidrug-resistant with a few of the isolates defined as extremely resistant (resistant to eight or more antimicrobial classes 39 ); thus, innovative nonantibiotic strategies are needed to limit Salmonella colonization of food animals and contamination of their food products 40 . Despite the availability of over 19,000 antimicrobial agents 41 , the literature on MDR Salmonella www.nature.com/scientificreports/ susceptibility to AMPs is notably sparse 42 . AMPs are ideal molecules to test against MDR Salmonella because AMPs generally induce rapid bacterial killing, have a broad spectrum of antimicrobial activity, and have a reduced proclivity for cultivating bacterial resistance compared to classical antibiotics 10 . It has previously been reported that 30-mer peptides derived from bovine NK-lysins are potent inhibitors against a variety of bacteria 17,20,21,[23][24][25] . This study established NK2A and NK2B as highly effective against Salmonella isolates (MDR and non-MDR). Both peptides were able to inhibit the growth of Salmonella at low micromolar concentrations (3.5-5 µM) during a 24 h incubation in LB broth (Fig. 1a). Although certain AMPs show reduced antimicrobial activity in the presence of salt 12,13,43,44 , NK2A and NK2B were highly effective in both LB broth and PBS containing physiological concentrations of salt (NaCl ~ 100-150 mM). www.nature.com/scientificreports/ The antimicrobial activity of cationic AMPs is predicted to be due to nonspecific interaction with bacterial membranes 7 . Bovine NK-lysin peptides are cationic α-helical AMPs and are expected to electrostatically interact with negatively charged bacterial membranes, resulting in pore formation, intracellular content leakage and bacterial death 20,21 . The net positive charges of bovine NK-lysin peptides in physiological buffer (PBS) is predicted to be + 5.0 to + 7.9. In general, bacterial surface charge or Zeta potential is negative due to the presence of negatively charged lipopolysaccharides in Gram negative bacteria and teichoic acids in Gram positive bacteria. Due to the complex nature of Gram negative bacterial LPS molecules, interactions of AMPs and LPS are poorly understood. Phosphatidylethanolamine, phosphatidylglycerol and cardiolipin are the major lipid components of bacterial cytoplasmic membranes among which the latter two are negatively charged 45 . Using a Zetasizer, it was confirmed that NK-lysin peptides have net positive Zeta potentials while Salmonella isolates have net negative Zeta potentials (Tables 2-3). Incubation of Salmonella isolates with NK2A resulted in increased bacterial Zeta potentials, suggesting electrostatic interaction and charge neutralization between NK2A and bacterial LPS and cytoplasmic membrane. Some studies suggested that increased Zeta potentials can be used as a potential marker to assess damage to bacterial membranes or viability 46 . However, increased Zeta potentials of Salmonella (isolate SB395) were observed even with Sc-NK2A which has minimal bacterial killing activity. Therefore, the findings suggest that increased bacterial Zeta potential alone is an imperfect tool for assessing AMP-mediated membrane permeability of Salmonella isolates.
Although direct bovine NK-lysin peptide binding to bacterial membranes has not been demonstrated, the increase in Zeta potential of Salmonella with NK2A in this study suggested NK2A-membrane interactions. Membranolytic activity of NK2A is supported by our experimental findings from live/dead bacterial viability staining, PI uptake assay, and TEM. Moreover, two specific experiments were conducted to confirm NK2A binding with MDR Salmonella SX238 membranes. Preincubation of NK2A with increasing concentrations of Salmonella LPS prevented NK2A-mediated MDR Salmonella SX238 growth inhibition (Fig. 6a). This suggests that NK2A-LPS binding is irreversible. TEM detection of NK2A-AuNP conjugates on both bacterial outer-and inner-membranes confirmed direct binding of NK2A to MDR Salmonella SX238 LPS and the cytoplasmic membrane.
Our ultimate goal is to develop non-antibiotic strategies to control Salmonella colonization and shedding in food animals. Of the four bovine NK-lysin peptides tested in this study, NK2A appeared to be the best candidate. In general, AMPs show poor pharmacokinetic properties because of their small size, hydrophobicity, and sensitivity to proteolytic degradation. Polyanhydride nanoparticles have been extensively characterized as a delivery platform for small molecules (including antimicrobials) and biomolecules 29,30,32,[47][48][49][50][51] . Polyanhydride nanoparticles entrap and protect sensitive payloads from harsh physiological environments (like proteases and low pH), traffic the payload to the site of action, and control the release of the payload 27,28 . These features could improve peptide stability, enhance AMP bioavailability, reduce proteolytic degradation, and decrease treatment frequency. In vitro release kinetics assays revealed that all tested nanoparticle formulations produced slow and sustained release of NK2A. These nanoparticle formulations also preserved NK2A functional activity, given that ~ 80% (in the case of 20:80 CPTEG:CPH particles ) to 100% (for 10:90 CPTEG:SA and 20:80 CPH:SA particles) MDR Salmonella killing was observed. These findings suggest that polyanhydride nanoparticles have the features of a suitable nano-carrier delivery platform for NK2A, and merit further investigation in future animal trials.
In conclusion, this study revealed that among four bovine NK-lysin peptides, NK2A and NK2B showed potent antimicrobial activity against Salmonella isolates (MDR and non-MDR). Using a variety of techniques, it was demonstrated that NK2A directly and irreversibly binds to bacterial LPS and the cytoplasmic membrane, leading to membrane permeability, intracellular content leakage, and bacterial death. Initial investigation of NK2A-loaded polyanhydride nanoparticles indicates both sustained release and peptide protection capabilities, motivating further exploration of a nano-carrier delivery approach for NK-lysin-based therapies to treat MDR Salmonella infections. Such an innovative, non-antibiotic strategy could aid efforts to control the spread of MDR Salmonella infections in animal and human populations.

Peptide concentration determination.
Concentrations of all NK-lysin peptides and Sc-NK2A were determined by quantitative amino acid analysis at the Protein Chemistry Laboratory, Texas A&M University and the Center for Biotechnology, University of Nebraska-Lincoln.
Animals and ethics. The authors complied with the ARRIVE guidelines. All animal procedures were conducted in strict accordance with the Animal Welfare Act, and all procedures were approved by the Institutional Animal Care and Use Committee at the National Animal Disease Center, Ames, IA. Blood samples collected from 4 years old healthy two Holstein cows from National Animal Disease Center were used to isolate red blood cells. www.nature.com/scientificreports/ Mouse study. To determine whether NK-lysin peptides show in vivo cytotoxic effects, a mouse study was conducted. Briefly, thirty 5-6 weeks old female Swiss-Webster mice (Envigo, Indianapolis, IN) were housed in groups in six individual cages (5 mice per cage) in ABSL2 facilities with ad libitum access to food and water in a 12-h light:dark cycle. After 1 week of acclimation, mice were intraperitoneally injected with a single dose of 100 µg of NK-lysin peptides (NK1, NK2A, NK2B, NK2C, or Sc-NK2A, ~ 4 mg/kg of body weight) in PBS (1 mL). Mice injected with PBS (1 mL) served as a negative control for the study. Mice were daily observed for any signs of toxicity and were euthanized 2 weeks post peptide administration.
Circular dichroism (CD) analysis. CD analysis of all four bovine NK-lysin peptides were previously described 24 . Therefore, CD analyses were performed here using the same methods for NK2A and Sc-NK2A in potassium phosphate buffer and liposomes using Jasco J-815 CD spectrophotometer (Jasco, Easton, MA).
Hemolytic activity assay. The hemolytic activity of all four bovine NK-lysin peptides, as the amount of hemoglobin released by the lysis of cattle red blood cells (RBCs), were previously determined 20 . Therefore, only hemolytic activity of Sc-NK2A was compared with NK2A (final concentrations for both peptides at 2, 5, 10, and 20 µM) using cattle RBCs in this study as described previously 20 . Triton X-100 and PBS were used as positive and negative control for the assay, respectively.
Nanoparticle synthesis. SA  MDR Salmonella-NK-lysin peptide susceptibility assays. Overnight cultures of a MDR Salmonella outbreak isolate (SX238) and one non-MDR Salmonella isolate (SB377) were diluted 1:1000 in LB broth to obtain approximately 1 × 10 6 colony forming units per milliliter (CFU/mL). One hundred microliters of each diluted culture were transferred into honeycomb® 100-well plates (~ 1 × 10 5 CFU/well) (Growth Curves USA, Piscataway, NJ). Bovine NK-lysin peptides (NK1, NK2A, NK2B, and NK2C) were added at the final peptide concentrations of 20, 10, 5, 3.5, 2 µM in PBS (total volume = 20 µL) with a final bacterial-peptide volume of 120 µL/well. The plates were covered with lids and placed in an automated growth curve reader (Bioscreen C; Growth Curves USA, Piscataway, NJ) which was programmed for continuous shaking at 37 °C, and optical density readings at 600 nm (OD 600 ) were recorded every hour for 24 h. Bacteria incubated with kanamycin and PBS were used as positive and negative control for the assay, respectively. The optical density data were analyzed, and growth curve graphs for each isolate were generated using MS Excel. Because NK2A peptide showed the strongest Salmonella growth inhibition, this peptide was selected for further studies. Twenty-four-hour growth curve assays for Salmonella isolate SX238 were repeated at different bacterial CFU (~ 1 × 10 5 , 1 × 10 6 , 1 × 10 7 , and 1 × 10 8 CFU/mL) with NK2A (5 µM and 10 µM). Sc-NK2A was included as a control for the assay. The bactericidal effect of NK2A peptide against MDR Salmonella (SX238) was also determined. Overnight SX238 culture was centrifuged (10,000 rpm for 2 min) and diluted 1:1000 in LB broth and/or PBS to obtain ~ 1 × 10 6 CFU/mL; 100 µL of diluted bacterial culture was placed in a Honeycomb plate (~ 1 × 10 5 CFU/ well), NK2A (20, 5, 2 and 0 µM) was added and incubated at 37 °C in the reader with continuous shaking for 3 h. Bacteria in each well was 10 × serially diluted in LB broth, 10 µL was plated on Xylose-Lysine-Tergitol 4 (XLT4) (ThermoFisher) agar plates and incubated at 37 °C overnight. Bacterial colonies were enumerated for each NK2A dilution.
Antimicrobial activity of NK2A (final concentration ~ 5 µM) encapsulated nanoparticles (CPTEG:CPH, CPH:SA and CPTEG:SA) were tested against Salmonella isolate SX238 as previously described 32  www.nature.com/scientificreports/ sonicated as previously described 32 . One hundred microliters of NK2A nanoparticles and 100 µL of diluted bacterial culture in PBS was added into a Honeycomb plate (~ 1 × 10 5 CFU/well) and incubated at 37 °C in the reader with continuous shaking for 3 h. Bacteria in each well was 10 × serially diluted in LB broth, 10 µL was plated on XLT4 agar plates and incubated at 37 °C overnight. Kanamycin (final concentration = 16 µg/mL) encapsulated and empty nanoparticles were used as positive and negative control for the assay, respectively. Bacterial colonies were enumerated for each nanoparticle treatment. The percentage of bacterial killing was calculated using the following Eq. (1 − [average number of live bacteria in NK2A-treated sample/average number of live bacteria in the negative control sample] × 100). The antimicrobial assays for Salmonella isolates were repeated at least three times in duplicate wells.
Zeta potential measurements. Salmonella isolates were prepared for Zeta potential measurements via the diffusion barrier technique as previously described but with some minor modifications 35 . Briefly, overnight cultures of Salmonella in LB broth were centrifuged (10,000 rpm for 2 min), washed twice in Milli-Q water and resuspended in Milli-Q water (~ 1 × 10 9 CFU/mL). Bacteria were then diluted 1:1000 in Milli-Q water supplemented with 10% PBS, then transferred to clear disposable folded capillary Zeta cells filled with 10% PBS via gel loading pipette tip (Malvern Instruments Ltd). The Zeta potential measurements (mV) of bacterial isolates were immediately measured using a Zeta potential analyzer equipped with He-Ne 633 nm laser (Zetasizer Nano, Malvern) at a constant voltage of 150 V at room temperature. The Zeta potential changes of Salmonella isolates following incubation with NK2A were also recorded. Bacterial isolates (~ 10 6 CFU in 1 mL Milli-Q water) were incubated with 2 µM NK2A (in PBS) at 37 °C for 20 min. Capillary Zeta cells were washed with Milli-Q water between samples and refilled with Milli-Q water containing 10% PBS before measuring the Zeta potentials. Zeta potential changes of one Salmonella isolate (SB395) was also similarly tested with Sc-NK2A (2 µM) peptide. This method was also used for assessing NK-lysin peptide Zeta potentials. A total of five Zeta potential measurements of 40 runs each with 30 s cooling cycles for each sample was performed.

Confocal laser scanning microscopy (CLSM). The viabilities of bacteria following incubation with
NK2A and Sc-NK2A were determined using culture-independent (function of the membrane integrity of the bacteria) staining technique by using LIVE/DEAD™ BacLight™ bacterial viability kit (for microscopy) as described previously 20 . Bacteria in each treatment was visualized using a Nikon A1R + laser scanning confocal microscope (Nikon Instruments, Melville, NY) as described previously 20 . Images were obtained using plan Apo 60 × objective lens (oil) at numerical aperture 1.4. Final figures were prepared using Adobe Photoshop Elements 11.

Transmission electron microscopy (TEM). Morphological changes of bacteria following incubation
with NK2A and Sc-NK2A peptides were studied by TEM. An overnight culture of Salmonella (isolate SX238) was centrifuged and resuspended in PBS. One hundred microliters of bacterial suspension were placed in a Honeycomb plate (~ 1 × 10 7 CFU/well) and NK2A, Sc-NK2A (20 µM), and PBS were added and incubated at 37 °C with continuous shaking for 1 h. The bacterial suspensions were mixed with an equal volume of 3% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) and processed for TEM as described before 24 . Sections were examined with a FEI Tecnai G2 Biotwin (FEI Co., Hillboro, OR) TEM and images were taken with Advanced Microscopy Technologies (AMT Inc., Danvers, MN) imaging camera.
Propidium iodide (PI) uptake assay. The kinetics of NK2A-induced damage to Salmonella cell membranes were evaluated using PI uptake assay as described previously 24 . Briefly, an overnight culture of MDR Salmonella outbreak isolate SX238 was centrifuged, washed with PBS, and resuspended in PBS. Ninety microliters of bacterial suspension was transferred to a 96-well opaque plate (~ 1 × 10 7 CFU/well), mixed with 10 µL of PI (final concentration = 3 µM) and incubated at 37 °C for 5 min. PBS, polymyxin B (2, 4, 25, 50 and 100 µg/mL), NK2A or Sc-NK2A (both at 2 and 5 µM) were added and PI fluorescent intensities in each well was continuously measured (15 s intervals, flex mode) using a FlexStation 3 multi-mode (fluorescent) microplate reader (Molecular Devices LLC., San Jose, CA) for up to 6 h. The excitation and emission wavelengths for PI was set up at 530 nm and 620 nm, respectively.

NK2A-Lipopolysccharides (LPS) binding studies.
The binding of NK2A to Salmonella LPS was studied by growth curve assay and TEM. Five micromolar NK2A (~ 2.2 µg/well) was preincubated with different amounts of Salmonella enterica serotype Typhimurium LPS (Sigma; LPS = 0, 0.5, 1.1, 2.2, 4.4, 8.8 µg/well, final volume of LPS and NK2A = 20 µL) at 37 °C in a honeycomb plate with continuous shaking for 30 min. Overnight Salmonella (isolate SX238) culture was diluted to ~ 1 × 10 6 CFU/mL and 100 µL was transferred into wells containing NK2A-LPS and further incubated at 37 °C with continuous shaking for 24 h. The optical density (OD 600 ) data were analyzed, and growth curve graphs were generated.
The direct binding of NK2A to Salmonella LPS/membranes was also examined by TEM using gold nanoparticle-conjugated NK2A (NK2A-AuNPs). Ten nanometer (diameter) NHS-activated gold nanoparticles (AuNPs) were conjugated to NK2A as described by the manufacturer (Cytodiagnostics Inc., Burlington, ON, Canada). One hundred microliters of Salmonella isolate SX238 was placed in a Honeycomb plate (~ 1 × 10 7 CFU/well) and incubated with NK2A-AuNPs (20 µM) or AuNPs (without NK2A) at 37 °C with continuous shaking for 15-45 min. Bacterial samples were fixed with 3% glutaraldehyde and processed for TEM as described above.