Physicochemical-guided design of cathelicidin-derived peptides generates membrane active variants with therapeutic potential

The spread of multi-drug resistance and the slow pace at which antibiotics come onto the market are undermining our ability to treat human infections, leading to high mortality rates. Aiming to overcome this global crisis, antimicrobial peptides are considered promising alternatives to counter bacterial infections with multi-drug resistant bacteria. The cathelicidins comprise a well-studied class of AMPs whose members have been used as model molecules for sequence modifications, aiming at enhanced biological activities and stability, along with reduced toxic effects on mammalian cells. Here, we describe the antimicrobial activities, modes of action and structural characterization of two novel cathelicidin-like peptides, named BotrAMP14 and CrotAMP14, which were re-designed from snake batroxicidin and crotalicidin, respectively. BotrAMP14 and CrotAMP14 showed broad-spectrum antibacterial activity against susceptible microorganisms and clinical isolates with minimal inhibitory concentrations ranging from 2–35.1 μM. Moreover, both peptides had low cytotoxicity against Caco-2 cells in vitro. In addition, in vivo toxicity against Galleria mellonella moth larvae revealed that both peptides led to>76% larval survival after 144 h. Microscopy studies suggest that BotrAMP14 and CrotAMP14 destabilize E. coli membranes. Furthermore, circular dichroism and molecular dynamics simulations indicate that, in a membrane-like environment, both peptides adopt α-helical structures that interact with bilayer phospholipids through hydrogen bonds and electrostatic interaction. Thus, we concluded that BotrAMP14 and CrotAMP14 are helical membrane active peptides, with similar antibacterial properties but lower cytotoxicity than the larger parent peptides batroxicidin and crotalicidin, having advantages for drug development strategies.

The discovery of penicillin and streptomycin during the golden age of antibiotics heralded a revolution in medical treatment of infections. Nevertheless, the spread of antibiotic resistance has made these antibiotics ineffective against many common pathogens 1 . The overuse and/or misuse of antibiotics results in microbial resistance through a variety of mechanisms 1 . Horizontal transfer of resistance determinants within and between species, and /or DNA mutations, has led to the emergence of pathogens resistant to multiple drugs 2 , reducing our capacity to treat multi-drug resistant bacterial infections 1 . Consequently, research on the development of new strategies to control antibiotic-resistance has intensified, including the usage of antimicrobial peptides (AMPs), which represent one of the oldest innate defense mechanisms in living organisms [3][4][5] . The composition and biological effects peptide synthesis and preparation. The peptides were purchased from Peptides 2.0 (USA) at a purity of >95%. The molecular mass for the peptides was confirmed using Matrix Assisted Laser Desorption Ionization -Time of Flight (MALDI-ToF) on a mass spectrometer Ultraflex MALDI-TOF III (Bruker Daltonics) 24 . The concentration of the designed peptides was determined using measurements of absorbance at 205, 215 and 225 nm as described by Niebergall and co-workers 25 . In vitro antimicrobial assays. Minimal inhibitory concentration (MIC) and minimal bactericidal concentration (MBC) experiments were performed for the ATCC and Clinical Isolated strains of E. coli 25922 K. pneumoniae 13822, and S. aureus 25923. The antimicrobial activities of all the peptides were tested as previously described with minor modifications 26 . Briefly, Mueller-Hinton (Himedia) broth was used to grow the strains overnight at 37 °C. MIC measurements were performed using 1 × 10 5 CFU.mL −1 and serial dilution of the peptides BotrAMP14 and CrotAMP14 starting at 50 µM. The MIC (100% inhibitory concentration) was determined after 24 h of incubation at 37 °C. The absorbance was measured in a 96-well microplate at 600 nm. Bacteria cultured only in MHB and containing the antibiotics (chloramphenicol, gentamicin and imipenem) were used as negative and positive control, respectively. MBCs were determined by plating out 10 µL of the contents of the wells where no bacterial growth was observed on MH agar plates. MBC was recorded as the lowest concentration at which no colonies were observed after 24 h incubation at 37 °C. All the measurements were performed in triplicate. neutral-red (nR) in vitro toxicity assay. Toxicity of the peptides on Caco-2 cells, at increasing peptide concentrations, was determined using a neutral-red (NR) assay as described previously 27 . After overnight incubation of Caco-2 cells with the peptides BotrAMP14 and CrotAMP14 at concentrations from 2-35 µM, 10 μL of a 33 µg.mL −1 NR solution was added to the wells containing the peptide-incubated cells. After 3 h of incubation, the supernatant was removed, and the cells were washed with phosphate buffer saline (PBS). Next, 150 μL of 1% acetic www.nature.com/scientificreports www.nature.com/scientificreports/ acid-50% ethanol was added and shaken for 10 min at room temperature. Finally, the absorbance was measured at 540 nm and 690 nm (background absorbance) using a SpectraMax M5 microplate reader (Molecular Devices). Readings were expressed as NR uptake relative to the uptake of the cells exposed to the negative control (medium or DMSO). All the measurements were performed in triplicate.
Galleria mellonella in vivo toxicity assay. In vivo toxicity was assayed using Galleria mellonella larvae 28 in their final instar stage. The larvae were purchased (UK Waxworms Ltd, Sheffield, UK), and acclimatized in the dark at 15 °C, and used within 14 days. Only larvae weighing between 0.2 and 0.3 g were used for experiments. Larvae were injected with 20 μL of BotrAMP14 and CrotAMP14 peptide solutions, or controls in the left posterior proleg using Terumo Myjector 29 G insulin syringes (VWR International). Two negative control groups were included in every experiment; one group was not injected to control for background larval mortality (no manipulation control) and the other group (uninfected control) was injected with PBS to control for the possible effect of physical trauma on mortality. After injection, larvae were acclimatized in Petri dishes in the dark at 37 °C with 5% CO 2 for up to 144 h post-inoculation and inspected every 24 h for survival. For each sample (non-manipulated control, water control, peptides) fifteen randomly chosen larvae were used. The peptide concentration was 10 mg. kg −1 of body weight. The data were expressed as % larval survival of the survival of the uninfected control. All the measurements were performed in triplicate. circular Dichroism. Circular dichroism (CD) spectra were obtained using a Jasco-715 spectro-polarimeter equipped with a Peltier element for temperature control. CD spectra were obtained in the far-UV range, from 200 to 260 nm using a quartz cuvette with 0.1 cm optical path. An averaging of 20 single scans was obtained for the peptide concentration of 110 μM, for demi water, 10 mM potassium phosphate buffered saline at pH 7.4, and 30 mM of sodium dodecyl sulphate (SDS). All the spectra were fitted using the CONTIN algorithm as implemented in the DICHROWEB 29-32 webserver, using the data basis set #7 for a quantitative interpretation of the spectra in terms of a percentage of α-helical structure. Molecular modeling. Initially, BLASTp 33 was performed in order to find the best template structures for the molecular modeling of BrotAMP and CrotAMP. Further, 100 theoretical three-dimensional models were constructed using Modeller v. 9.12 33 , based on the nuclear magnetic resonance (NMR) structure of a crotalicidin isolated from venom of the rattlesnake (Crotallus durissus) (PDB code: 2MWT) 18 . The lowest free-energy theoretical models (DOPE score) for both peptides were then selected and used for validation procedures according to their fold (ProSA-web) 34 and stereochemistry (PROCHECK) 35 . Structure visualization was performed in PyMOL (http://www.pymol.org).

Molecular dynamics in water and SDS micelles. Molecular dynamics (MD) simulations for
BotrAMP14 and CrotAMP14 were initially carried out in contact with an SDS micelle, according to Cardoso et al. 36 . The simulations were performed using the CHARMM36 force field from the computational package GROMACS v.5.0.4 37 . MD simulations in SDS were carried out in dodecahedron boxes, where both peptides were put in contact with SDS micelles constituted of 100 detergents. SDS micelles were built, and their topologies generated using the CHARM-GUI server 24 . Chloride ions were added to neutralize the systems' charge in both simulations. The simulations were performed under 0.15 M NaCl ionic strength. Geometry of water molecules was constrained using the SETTLE algorithm 38 . Moreover, the LINCS algorithm 39 was used to link all the atom bond lengths. Particle Mesh Ewald (PME) 40 was used for electrostatic corrections with a radius cut-off of 1.4 nm to minimize the computational simulation time. The same radius cut-off was used for van der Waals interactions. The list of neighbors of each atom was updated every 10 simulation steps of 2 fs each. The steepest descent algorithm (50,000 steps) was applied for energy minimization. The systems underwent a normalization of temperature and pressure to 310 K and 1 bar using the velocity rescaling thermostat (NVT) 41  Scanning electronic microscope. Scanning electronic microscopy (SEM) was performed by the adherence of bacteria to Poly-L-Lysine-coated slides. Prior to SEM analysis, E. coli ATCC 25922 was grown in MHB overnight at 37 °C, and the concentration of 1 × 10 5 CFU.mL −1 . The bacterial cultures were centrifuged, and phosphate buffer was used to replace the MHB. The bacterial culture was increased by adding 8 µM and 2 µM for BotrAMP14 and CrotAMP14, respectively, for 5, 30 and 60 min of incubation. For control group, E.coli ATCC 25922 was used without peptides (time 0). Samples for SEM were prepared by leaving microscope slides coated with poly-L-lysine in 10 mL suspension of bacteria incubated with the peptides and allowing them to settle and adhere to the slides. After 60 min incubation at RT, the bacteria were fixed using 2.5% glutaraldehyde in phosphate buffer. Finally, water was removed using critical point drying: first, the samples were immersed in a graded series of increasing ethanol: 25%, 50%, 75%, and two times at 100% each for 10 min. This was followed by transferring the samples to absolute ethanol, and by critical point drying.
Scientific RepoRtS | (2020) 10:9127 | https://doi.org/10.1038/s41598-020-66164-w www.nature.com/scientificreports www.nature.com/scientificreports/ Cationic AMPs from several classes (cathelicidins, defensins and magainins), are commonly modified using rational design strategies, which allow a shorter motif to be obtained, decreasing the cost of producing synthetic peptides for therapeutic applications 43 . In addition, the removal of specific residues with undesired properties may also produce peptide analogues with lower toxicity and immunostimulatory activities 43 . In the present work, the amino acid Lys (K 12 ) was inserted in both redesigned variants (BotrAMP14 and CrotAMP14) with the aim of boosting electrostatic interactions with bacterial membranes and promoting antibacterial activity. Furthermore, studies have shown that lysine-rich peptides have reduced toxicity against eukaryotic cells 46 . The Trp (W 3 ) was inserted in BotrAMP14 because of its ability to arrange itself more deeply into the bacterial membranes, presenting a distinct preference for the interfacial region of lipid bilayers 47 . Therefore we expected BotrAMP14 to disrupt the bacterial cell membranes more efficiently 47 . Overall, these amino acids modifications interfered at the helical wheel diagram for BotrAMP14 and CrotAMP14, resulting in clear cationic and hydrophobic residues distribution along the peptide sequence, thus favoring the amphipathicity (Table 1 and Fig. 1B).
The efficient antibacterial activity described here for the designed peptides BotrAMP14 and CrotAMP14 is consistent with previous works regarding the precursor CRAMPs 18,44,45 . Regarding antimicrobial activity, the higher MICs observed for the Gram-positive bacteria than for the Gram-negative bacteria tested indicate that both redesigned peptides possess a differential affinity for these types of bacterial membranes. By reducing the number of amino acid residues and increasing hydrophobic moment, the redesigned peptides had a higher MIC for the Gram-positive bacterium S. aureus than Gram-negative bacteria. This feature may be related to the increased helicity of the redesigned peptides BotrAMP14 and CrotAMP14 compared to the parental peptides. The propensity to form structures in amphipathic α-helices in membrane-mimicking membrane environments has been demonstrated in several studies as an essential factor for the disruptive activity of AMPs 48 . More recently, a smaller derivative of cathelicidin KP36, designated AMP (RN15), also demonstrated promising activity, mostly against Bacillus bacteria, and low hemolytic and cytotoxicity against dermal human dermal fibroblasts 49 . However, for Gram-negative strains, a slight increase of BotrAMP14 over the parental Btn [16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] was observed. This increase may be related to the tryptophan residue presence, which could lead to an increase in membrane/peptide interaction due to tryptophan insertion into the interfacial region of the phospholipid bilayer 47 . This increase was not observed for CrotAMP14 peptide, which maintained the overall parental activity. Ultimately, the bacterial cytoplasmic membrane lipid composition seems essential for AMPs' mechanism of action. Generally, it is constituted by zwitterionic lipid phosphatidylethanolamine (PE), and anionic lipids phosphatidylglycerol (PG) and cardiolipin (CL), which are essential for membrane organization. The lipid composition differs from one species to another. Usually Gram-positive bacteria present a high anionic lipid (PG and CL) amount, which may favor electrostatic interactions. Gram-negative bacteria show higher PE content in comparison to Gram-positive strains 50 . Such features could explain the higher increased activity against Gram-positive bacteria for the redesigned peptides over parental.

Structural analyses.
To elucidate the secondary structure of BotrAMP14 and CrotAMP14, CD spectroscopy, molecular modeling and dynamics simulations were carried out under different experimental conditions. CD spectra were acquired in demi-water, 10 mM potassium phosphate buffer at pH 7.4 and in 30 mM SDS, and they are shown in Fig. 2. A qualitative comparison with typical CD spectra illustrates that BotrAMP14 and CrotAMP14 spectra are characteristic of predominantly random coil configurations in demi-water and in 10 mM potassium phosphate buffer ( Fig. 2A,B). In contrast, CD spectra of BotrAMP14 and CrotAMP14 indicated substantial α-helical configuration in a membrane-like environment (SDS) (Fig. 2C). For a more quantitative assessment, we used the CONTIN algorithm to fit the CD spectrum and extract percentages of the α-helical secondary  (2A-B). The results show BotrAMP14 has an estimated percentages of random coil structure in demi water and 10 mM potassium phosphate buffer of ~38% and ~61%, respectively.
For CrotAMP14, the estimated percentages of random coil were very similar in demi-water and buffer (i.e. 48% and ~66% respectively). However, the dominant α-helical content of secondary structure in 30 mM SDS was ~65% and ~69%, respectively, for BotrAMP14 and CrotAMP14. Similarly, Chen and coworkers 51 also demonstrated that the smaller oligopeptide of 15 amino acid residues ( 1 VKRFKKFFRKLKKSV 15 ) derived from the cathelicidin BF-30 (Bf-CRAMP), revealing that an oligopeptide of 15 amino acid residues ( 1 VKRFKKFFRKLKKSV 15 ) maintained its minimal α-helical structure required for an antimicrobial activity. Another study on cathelicidins focused on the use of short synthetic peptides of sequences incorporated into Ophiophagus hannah peptides (Oh-CRAMP and Oh-CATH) and distinct antimicrobial activity and hemolysis were demonstrated relative to human erythrocytes in comparison with their original parent sequences 52 .
MD simulations were performed for similar conditions as those used in the CD experiments. MD simulations in the presence of SDS micelles showed that the SDS clearly promoted α-helical conformations for both BotrAM14 and CrotAMP14 (Fig. 3C,D). Also, supporting the results of the CD experiments, the two peptides differ in overall helical chemical contents (Fig. 3A,B). According to the RMSD analysis CrotAMP14 showed ~3-fold higher deviations (~0.45 nm) in its trajectory when compared to BotrAMP14 (~0.15 nm) (Fig. 3A). This was also reflected in the fluctuation per residue observed for these peptides, where higher RMSF values were observed for CrotAMP14 (Fig. 3B).
Our SEM results suggest that both BotrAMP14 and CrotAMP14 directly attach to bacterial surfaces and trigger membrane perturbation events in Gram-negative bacteria. Interestingly, as for our simulations in the presence of SDS micelles, CrotAMP14 deviated and fluctuated more during the entire simulation (Fig. 3A). A total of 16 atomic interactions were predicted for the molecular complex BotrAMP14/SDS. These interactions involved the residues K 1 , R 2 , W 3 , K 4 , K 5 , F 6 , R 8 , V 10 , I 11 , K 12 and F 14 in BotrAMP14 (Supplementary Table 1). Furthermore, 13 interactions were predicted for the complex CrotAMP14/SDS. These interactions involve the residues K 1 , R 2 , L 3 , K 4 , K 5 , I 6 , F 7 , K 12 and I 13 in CrotAMP14 (Supplementary Table 2). These data clearly indicate the relevance of the positively charged residues of both peptides for initial electrostatic and hydrogen bond interactions with the sulfate groups of the SDS molecules.

Cytotoxicity of BotrAMP14 and CrotAMP14.
To be used as drugs, antibacterial compounds must have high selectivity to the pathogens at therapeutic concentrations in the body and must lack cytotoxicity to the host cells. Usually, the cytotoxic properties associated with both naturally occurring and rationally designed AMPs represent a bottleneck in the treatment of infections using these molecules. In this context, the toxic effects of BotrAMP14 and CrotAMP14 on mammalian cells were initially investigated in vitro using Caco-2 cells. Neutral red uptake (NRU) assay was used to determine the viability of mammalian Caco-2 cells as a function of peptide concentration. We observed that cells incubated with BotrAMP14 showed 100% viability at the maximum concentration tested (88 μM (Fig. 5A). At this same concentration, however, the CrotAMP14 peptide was cytotoxic, reducing cell viability to 60% (Fig. 5A). Caco-2 cells can be widely used to test the intestinal permeability and toxicity for several drugs 54 . The lack of toxicity of BotrAMP14 and CrotAMP14 against Caco-2 cells is very promising and can be used for future intestinal permeability assays, since antibiotics may lead to an imbalance of the microbiota, which is important when considering the inclusion of new drugs on the market. www.nature.com/scientificreports www.nature.com/scientificreports/ Toxicity of BotrAMP14 and CrotAMP14 to Galleria mellonela larvae. Infectivity trials and toxicity testing in rodents are important requisites for the use of drug candidates in humans. However, trials in rats and mice are expensive and there are ethical considerations. G. mellonella (greater wax moth) larvae are a potential alternative as they have evolutionarily conserved immunity consisting of both cellular and humoral defenses 55 .
In vivo experiments were performed using greater wax moth G. mellonella larvae for a period of 144 h, at a peptide concentration of 10 mg.kg −1 of body weight (Fig. 5B). Survival curves indicated that both BotrAMP14 and www.nature.com/scientificreports www.nature.com/scientificreports/ CrotAMP14 have low toxicity for the larvae. For BotrAMP14, we observed 86.6% of larval survival after 144 h of experiment. The peptide CrotAMP14 was slightly more toxic, resulting in 76.6% larval survival after 144 h (Fig. 5B). Latour and coworkers 21 showed the antimicrobial activity of peptides derived from Naja atra, composed of the ATRA motif, i.e, KR(F/A)KKFFKK(L/P)K, with a trivial toxicity against erythrocytes 21 . Contrary to what was demonstrated here, Wang and coworkers 56 showed that derivative crotalicidin EVP50 presented the highest toxicity activity against zebrafish larvae in comparison to the original vipericidin sequences. On the other hand, other studies performed with analogues/derivatives of cathelicidins reported that these peptides had low toxicity and hemolytic activity 51 .
In conclusion, it was possible to observe that the physicochemical-guided design of BotrAMP14 and CrotAMP14 preserved the antibacterial and non-toxic potential of their parent sequences despite their smaller size, suggesting the applicability of these variants as new drug candidates. We also concluded that both peptides act on Gram-negative bacteria through a membrane destabilization mechanism, which is mainly driven by electrostatic interactions and hydrogen bonding between the N-terminus region of these peptides and anionic membranes, leading to peptide anchoring and insertion. Overall, BotrAMP14 and CrotAMP14 appear as suitable candidates for further drug development, especially for the treatment of resistant Gram-negative bacteria-associated infections.