TB, TB_L1FK (designed as reported in “Results”) and TB_KKG6A were synthesized by Proteogenix (Schiltigheim, France). Analysis of the synthetic peptides by high performance chromatography (HPLC) and mass spectrometry revealed purity over 98%. Peptides were diluted in milli-Q water to obtain a stock solution of 1 mM and stored at −80°C. The main features of the peptides are shown in Table 1.

Disodium ethylenediaminetetraacetic acid (EDTA) was purchased from Sigma-Aldrich (St. Louis, USA). A stock solution of EDTA (0.5 M) was prepared in milli-Q water by adjusting the pH to 8.0 with NaOH. The working solution (50 mM) was obtained by diluting the stock solution in milli-Q water, sterile filtered and stored at 4°C.

The reference laboratory strains Klebsiella pneumoniae (ATCC BAA-1706), P. aeruginosa (ATCC 27853), S. aureus (ATCC 33591), and S. epidermidis (ATCC 35984) were used for the study. For the preparation of stock cultures, bacterial strains were grown in Tryptone Soy Broth (TSB) (Oxoid, Basingstoke, UK) until mid-log phase, subdivided in aliquots and stored at −80°C. For the colony-forming units (CFU) count, serially diluted bacterial suspensions were plated on Tryptone Soy Agar (TSA) (Oxoid) and incubated for 24 h at 37°C.

The bactericidal activity of TB, TB_L1FK and TB_KKG6A against K. pneumoniae, P. aeruginosa, S. aureus, and S. epidermidis was evaluated by the microdilution method in sodium-phosphate buffer (10 mM SPB, pH 7.4). Bacterial strains were grown in TSB until exponential phase and suspended in SPB to reach a density of 1 × 10⁷ CFU/mL. A volume of 10 μL of the bacterial suspensions was added to 90 μL of SPB containing different concentrations of the peptides (from 1.5 to 48 μM). Bacteria suspended in SPB alone were used as cell viability control. Samples were incubated at 37°C with shaking for various times (5, 15, 30, 60, and 90 min), subsequently diluted 10-fold in TSB and plated on TSA to determine the number of CFU. The minimal bactericidal concentration (MBC) was defined as the minimal concentration of peptide causing a reduction of at least 3 Log₁₀ in the number of viable bacteria after 90 min of incubation (Mangoni et al., 2008).

The ability of TB, TB_L1FK, and TB_KKG6A to prevent biofilm formation was evaluated against S. aureus and P. aeruginosa. Bacteria were grown overnight in TSB/Glc (TSB added with 0.25% (v/v) glucose) at 37°C. Stationary-phase cultures were diluted 1:1,000 in Biofilm Promoting Medium (BPM; TSB diluted 1:1 with 10 mM SPB at pH 7.4 and supplemented with 0.25% glucose). Bacterial suspensions were inoculated into flat-bottom polystyrene 96-well microplates (Corning Costar, Lowell, USA), in the absence (negative control) or in the presence of different concentrations of each peptide (from 12 to 48 μM). Microplates were incubated statically at 37°C for 24 h and biofilm biomass was estimated by crystal violet (CV) staining assay. To this aim, biofilms were rinsed three times with phosphate-buffer saline (PBS), air-dried for 15 min and incubated with 0.1% (w/v) CV (bioMérieux, Florence, Italy) for 15 min. The excess of CV was removed by washing the plates with PBS, while biofilm-associated CV was extracted with 98% ethanol (Sigma Aldrich) and quantified by measuring the optical density at 570 nm (OD₅₇₀) in a microplate reader (Model 550, Bio-Rad Laboratories Srl, Italy).

The activity of TB, TB_L1FK, and TB_KKG6A against preformed (24-h old) biofilms of S. aureus and P. aeruginosa was also investigated. Briefly, biofilms were allowed to form for 24 h in flat-bottom 96-well microplates in the absence of antimicrobial compounds. Established biofilms were then washed three times with PBS in order to remove non-adherent cells and incubated in fresh BPM with different concentrations of the three peptides (from 15 to 120 μM). After 24 h of incubation, the viability of biofilm-associated cells was evaluated by CFU counting. For this purpose, biofilms were washed three times with PBS and bacterial cells were detached from the surface of the wells with a pipette tip, vigorously vortexed and plated in serial dilutions on TSA.

TB_L1FK and TB_KKG6A were combined with EDTA in order to enhance their activity against preformed biofilms of S. aureus and P. aeruginosa. To this aim, 24 h-old biofilms of the two bacterial species were exposed to different concentrations of the peptides (15 and 30 μM), alone and in combination with EDTA (1.25 and 2.5 mM). Microplates were incubated statically at 37°C for 24 h. Following incubation, the antibiofilm effect was evaluated in terms of number of biofilm-associated viable cells as previously described.

The antibacterial activity of TB_L1FK and TB_KKG6A, used alone and in combination with EDTA, was also tested against planktonic cells of S. aureus and P. aeruginosa. The Minimal Inhibitory Concentration (MIC) of the peptides, EDTA and the peptide-EDTA combinations was determined by the microdilution method under the same experimental conditions used for the biofilm assay. Briefly, bacteria from overnight cultures were diluted 1:1,000 in BPM and incubated for 24 h at 37°C in propylene tubes in the presence of TB_L1FK and TB_KKG6A (from 3.75 to 120 μM), alone and combined with EDTA (from 0.3 to 10 mM). MIC was defined as the lowest concentration of the compounds resulting in the complete inhibition of visible growth. The effect of each combination on cell growth was studied using an adapted Fractional Inhibitory Concentration (FIC) index analysis. FIC index was calculated as follows: Σ (FIC_A + FIC_B), where FIC_A is the MIC of compound A in combination/MIC of compound A alone, and FIC_B is the MIC of compound B in combination/MIC of compound B alone. Synergism was defined as a FIC index ≤ 0.5, indifference as a FIC index > 0.5 and antagonism as a FIC index > 4 (Katragkou et al., 2015; Dosler et al., 2016).

Hemolytic activity of TB and its analogs was tested against human red blood cells (RBCs) as previously described (Tavanti et al., 2011). Briefly, peripheral blood obtained from healthy donors was centrifuged (1,000 × g for 10 min, 4°C) and washed three times with PBS (Euroclone, Milan, Italy). A suspension of RBCs (4%, v/v) was mixed with various concentrations of the peptides (from 12 to 96 μM) into a round-bottom polystyrene 96-well microplate (Corning Costar). RBCs suspended in PBS alone were used as negative control (0% hemolysis), while cells lysed with 0.1% Triton X-100 were taken as positive control (100% hemolysis). The microplate was incubated for 1 h at 37°C and then centrifuged at 1,000 × g for 20 min, 4°C. Supernatants were transferred to a new plate and the optical density at 450 nm (OD₄₅₀) was measured by means of a microplate reader. The hemolytic activity was quantified according to the following formula: hemolysis (%) = [(OD _peptide – OD _{negative control})/(OD _{positive control} – OD _{negative control})] × 100.

Cytotoxic activity of the peptides was assessed against human peripheral blood mononuclear cells (PBMCs) and human non-small-cell lung adenocarcinoma A549 cells (ATCC CCL-185).

PBMCs were isolated from buffy coats by conventional density gradient centrifugation. For this purpose, buffy coats were diluted 1:1 in PBS supplemented with 10% (v/v) sodium citrate (Sigma-Aldrich) and layered on Lympholyte-H gradient medium (Euroclone). Following centrifugation at 200 × g for 20 min at room temperature, the supernatant was eliminated in order to remove platelets. Buffy coats were further centrifuged at 800 × g for 20 min at room temperature and the lymphocyte/monocyte layer was harvested at the sample/medium interface. PBMCs were washed three times with PBS containing 0.5% (wt/v) bovine serum albumin (BSA; Sigma-Aldrich) and 10% sodium citrate, counted and re-suspended in RPMI 1640 (Euroclone) added with 10% (v/v) fetal calf serum (FCS; Euroclone) and 2 mM L-glutamine. Cells (1 × 10⁵ per well) were seeded into round-bottom 96-well microplates (Corning Costar) and incubated with increasing concentrations of the peptides (from 12 to 96 μM) for 24 h at 37°C, 5% CO₂. PBMCs incubated with culture medium were used as negative (cell viability) control, while cells treated with cycloheximide (2 mg/mL) served as a positive (death) control.

A549 cells were grown in tissue culture flasks in Dulbecco's modified Eagle's medium (DMEM; Euroclone) containing 10% FCS and 2 mM L-glutamine. Confluent monolayers of A549 cells were washed with PBS, treated with a trypsin-EDTA solution (Sigma-Aldrich), centrifuged at 300 × g for 10 min, counted and re-suspended in complete DMEM at a final density of 5 × 10⁴ cells/mL. A volume of 200 μL of the cell suspension was seeded into flat-bottom 96-well microplates (Corning Costar) and cultured for 24 h at 37°C, 5% CO₂. Peptides at a final concentration of 12–96 μM were added to the cells and incubated for further 24 h at 37°C, 5% CO₂. A549 cells incubated with culture medium were used as negative (cell viability) control, while cells treated with cycloheximide (2 mg/mL) served as a positive (death) control.

Cytotoxic activity was evaluated by the propidium iodide (PI) flow cytometric assay. To this end, PBMCs were washed once in PBS, resuspended in 100 μL, and incubated with 5 μL of a PI solution (50 μg/mL) (Sigma-Aldrich) for 4 min in the dark. Similarly, A549 cells were harvested by trypsinization, rinsed once with PBS and exposed to PI. Counting of viable (PI-negative) and dead (PI-positive) cells was carried out with a BD Accuri C6 flow cytometer (BD Biosciences, Mountain View, CA) and data were analyzed using BD Accuri C6 software (BD Biosciences). Cytotoxic effect was determined according to the following formula: Cytotoxicity (%) = [(PI-positive cells _peptide – PI-positive cells _{negative control})/(100 – PI-positive cells _{negative control})] × 100. The IC₅₀ values (Inhibitory Concentration) were defined as the concentration of the peptides causing 50% cell death as compared to the untreated control.

All the experiments were performed at least in triplicate, unless otherwise specified. Differences between mean values of groups were evaluated by one-way analysis of variance (ANOVA) followed by Tukey-Kramer post-hoc test, after normalization of the data. A p-value < 0.05 was considered statistically significant.

In order to improve the therapeutic potential of TB, a novel peptide was computationally designed starting from TB sequence. In a previous work, chemophysical analysis of known AMPs sequences was successfully employed to design a statistical model of membrane-disrupting peptides able to account for non-natural amino acids (Maccari et al., 2013). In this work, an additional statistical model was designed to account for peptides' cytotoxic effect. Together with the previously described models for the secondary structure and the antimicrobial activity, a forth constraint was imposed in order to retain as much as possible the sequence similarity with TB. A dataset of peptides with proved cytotoxic effect was appositely designed by collecting and combining data from different bioactive peptide databases (Gupta et al., 2013). Furthermore, another set of peptides was designed to represent non-cytotoxic peptides, allowing the statistical model to grasp the features that distinguish the two sets (see Section 1.1 in the Supplementary Material). A number of filters aimed to normalize and uniform the training set of peptides were applied and then, sequences were encoded into physicochemical variables representing global and topological properties of peptides. A machine learning algorithm was adopted to build a prediction engine able to discern between toxic and non-toxic peptides (see the Supplementary Material for details in model training and validation). Model performance was evaluated by the Mathews Correlation Coefficient (MCC), which assesses the prediction in terms of true and false positives and negatives. In the final configuration, a prediction model with an MCC value of 0.82 was obtained. The candidate sequence, named TB_L1FK, was designed by applying the statistical model to a particular class of Genetic Algorithms, called Multi-Objective Evolutional Algorithms (MOEA) that allows to screen for candidates that simultaneously satisfy different criterions.

As reported in Figure 1, that shows a predictive simulation of the structure of TB and its two analogs, TB_L1FK displays similar physicochemical characteristics to the parental peptide. Hydrophobicity and net charge of TB_L1FK are close to those of TB, while TB_KKG6A presents a different hydrophobic profile and an increased net charge, particularly localized at the C-terminus. One of the aims in the computational design of TB_L1FK was to retain all the features that could infer in the membrane interaction of the peptide with the target cells. Besides, molecular hydrophobicity and net charge, as well as size and molecular weight, represent important aspects for the loading and the controlled release of peptides such as TB from nanostructured delivery systems (Piras et al., 2015).

The antimicrobial activity of TB, TB_KKG6A, and TB_L1FK was evaluated in terms of MBC values toward S. aureus and S. epidermidis as models of Gram-positive bacteria and against K. pneumoniae and P. aeruginosa as models of Gram-negative bacteria. As shown in Table 2, TB was mainly active against Gram-positive bacteria and exhibited a bactericidal effect against Gram-negative bacteria only at 48 μM. Both analogs displayed a markedly increased activity compared to the parental peptide against all the bacterial species tested, but especially against the Gram-negative ones. In particular, a 2- to 8-fold reduction in the MBC compared to TB was observed against the Gram-positive bacteria, while an up to 16-fold decrease in the MBC value was observed in the case of the Gram-negative bacteria.

Time-kill studies on two representative bacterial species, S. aureus and P. aeruginosa, were carried out using the peptides at concentrations equal to their MBC. TB exerted its bactericidal activity toward S. aureus after approximately 90 min of incubation (Figure 2A). Both TB_L1FK and TB_KKG6A exhibited a faster killing kinetics than TB against the same bacterial species causing a reduction of at least 3 Log₁₀ in the number of viable bacteria within 30 and 60 min, respectively (Figure 2A). All three peptides showed a more rapid bactericidal effect against P. aeruginosa than against S. aureus (Figure 2B). In particular, TB and TB_KKG6A showed similar killing kinetics, being bactericidal after 15 min of incubation, while the most rapid bactericidal effect was exerted by TB_L1FK that determined the complete eradication of the starting bacterial inoculum within as little as 5 min of incubation (Figure 2B).

We first investigated the ability of TB, TB_L1FK and TB_KKG6A to inhibit the formation of biofilms of S. aureus and P. aeruginosa, two bacterial species often involved in the formation of biofilms particularly refractory to antimicrobial treatment. The inhibitory effect was assessed by CV staining (total biofilm biomass) evaluating the percentage of biofilm formation after 24 h of incubation with TB or the two TB analogs, as compared to the control biofilms (cells incubated in medium only). As shown in Figure 3A, differently from the parental peptide, TB_L1FK and TB_KKG6A reduced the ability of S. aureus to form biofilm of more than 50% as compared to the untreated control at 12 μM. All the peptides caused around 80% decrease of the biofilm biomass at the concentration of 24 μM. When the peptides were assayed against forming biofilms of P. aeruginosa, no inhibitory activity of TB and TB_L1FK was observed at concentrations up to 48 μM (Figure 3B). In contrast, TB_KKG6A displayed a considerable ability in reducing the biomass of P. aeruginosa biofilms, causing an 80% inhibition at the concentration of 24 μM (Figure 3B).

Secondly, the efficacy of TB and its analogs against preformed (24 h-old) biofilms of S. aureus and P. aeruginosa was evaluated by CFU counting after 24 h of incubation with the peptides. In the case of S. aureus biofilms, TB did not exert a considerable antibiofilm activity at concentrations up to 120 μM (data not shown), while TB_L1FK and TB_KKG6A caused a decrease of approximately 2 Log₁₀ in the number of biofilm-associated viable cells as compared to untreated biofilms at 30 μM (Figures 4A,B). When tested against biofilms of P. aeruginosa, none of the three peptides displayed a significant ability to reduce the number of CFU at the highest tested concentration (120 μM) (data not shown).

The possibility to improve the activity of TB analogs against preformed biofilms of S. aureus and P. aeruginosa was investigated combining the peptides with EDTA, a chelating agent previously reported to enhance the antibiofilm properties of TB (Maisetta et al., 2016). Indeed, the ability of EDTA to establish strong complexes with divalent cations essential for matrix stability could produce a matrix-disaggregating effect and promote the accessibility of peptides to biofilm-forming cells. The antibiofilm activity of various peptide-EDTA combinations was evaluated by CFU counting. Among all the tested combinations, the most powerful potentiating effect in terms of viable count reduction was obtained using both peptides at the concentration of 30 μM in combination with 1.25 mM (for S. aureus) or 2.5 mM EDTA (for P. aeruginosa). As regards S. aureus (Figures 4A,B), the combination of both TB_L1FK and TB_KKG6A with EDTA caused a reduction in the CFU number of approximately 1 Log₁₀ (90%) compared to the peptides and EDTA used alone, and 3 Log₁₀ (99.9%) compared to control biofilms after 24 h of incubation. Also in the case of P. aeruginosa, an enhanced ability of TB_L1FK and TB_KKG6A in biofilm reduction was demonstrated when peptides were used in combination with EDTA. Indeed, both peptide-EDTA combinations reduced the CFU number of approximately 1 Log₁₀ as compared to the peptide used alone (Figures 4C,D).

In order to investigate whether the synergism between TB analogs and EDTA was due to a disaggregating effect on biofilm extracellular matrix and/or to a direct effect on bacterial cells, we assessed the activity of the combination on planktonic bacteria in biofilm-like conditions (i.e., stationary phase cells suspended in BPM) in terms of MIC values. As shown in Table 3, when tested alone, TB_L1FK displayed MICs of 15 and 120 μM against S. aureus and P. aeruginosa, respectively. In the case of TB_KKG6A, the growth-inhibiting effect was recorded at 7.5 μM for S. aureus and at 30 μM for P. aeruginosa. In order to identify any synergistic interaction, sub-inhibitory concentrations of each peptide and EDTA were combined and the FIC index for the different peptide-EDTA combinations was calculated. Differently to what observed for the biofilm mode of growth, EDTA was not able to potentiate the antibacterial activity of TB_L1FK and TB_KKG6A against planktonic cells of S. aureus (FIC index > 0.5, Table 3). Conversely, a synergistic effect between both TB analogs and EDTA was observed against P. aeruginosa planktonic cultures (FIC index = 0.25, Table 3). Interestingly, the combination with EDTA produced an 8-fold decrease in the MIC of both peptides against planktonic P. aeruginosa grown in biofilm-like conditions, suggesting a direct effect of EDTA in displacing divalent cations that are required for the integrity of the outer membrane of Gram-negative bacteria (Gray and Wilkinson, 1965; Asbell and Eagon, 1966).

The hemolytic activity of TB and TB analogs was evaluated toward human RBCs. As shown in Figure 5, no hemolytic effect of the parental peptide was assessed at concentrations up to 96 μM. An overall increase in hemolytic activity of both analogs was observed. Nevertheless, a hemolysis below 10%, commonly recognized as a safe cut-off (Amin and Dannenfelser, 2006), was observed at concentrations up to 24 μM of TB_KKG6A and up to 48 μM of TB_L1FK.

TB, TB_L1FK and TB_KKG6A were tested for cytotoxic activity on PBMCs and A549 cells by flow cytometric determination of PI incorporation in cells treated with different concentrations of the three peptides. As shown in Figure 6, TB did not exhibit a significant cytotoxic effect toward both PBMCs and A549 cells at any of the tested concentrations. Indeed, an approximately 90% viability was observed at 96 μM for both cell types. Both TB analogs displayed higher cytotoxicity against both cell types as compared to TB (Figures 6A,B). When the toxic effect was evaluated as IC₅₀ value, TB_L1FK and TB_KKG6A showed comparable levels of cytotoxicity against PBMCs (IC₅₀ values of 52 and 49 μM, respectively). In contrast, TB_L1FK displayed lower levels of cytotoxicity against A549 cells with an IC₅₀ value of 59 vs. 16 μM of TB_KKG6A.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.