Structure-function-guided design of synthetic peptides with anti-infective activity derived from wasp venom

SUMMARY

Antimicrobial peptides (AMPs) derived from natural toxins and venoms offer a promising alternative source of antibiotics. Here, through structure-function-guided design, we convert two natural AMPs derived from the venom of the solitary eumenine wasp Eumenes micado into α-helical AMPs with reduced toxicity that kill Gram-negative bacteria in vitro and in a preclinical mouse model. To identify the sequence determinants conferring antimicrobial activity, an alanine scan screen and strategic single lysine substitutions are made to the amino acid sequence of these natural peptides. These efforts yield a total of 34 synthetic derivatives, including alanine substituted and lysine-substituted sequences with stabilized α-helical structures and increased net positive charge. The resulting lead synthetic peptides kill the Gram-negative pathogens Escherichia coli and Pseudomonas aeruginosa (PAO1 and PA14) by rapidly permeabilizing both their outer and cytoplasmic membranes, exhibit anti-infective efficacy in a mouse model by reducing bacterial loads by up to three orders of magnitude, and do not readily select for bacterial resistance.

Graphical Abstract

Antimicrobial peptides (AMPs) derived from natural venoms and toxins are a promising alternative to conventional antibiotics. Here, Boaro, Ageitos, et al. use a structure-function-guided design to convert two natural AMPs derived from the venom of the solitary eumenine wasp Eumenes micado into potent AMPs with reduced toxicity that kill Gram-negative bacteria in vitro and in a preclinical mouse model.

INTRODUCTION

Drug-resistant bacterial infections are a serious public health problem worldwide, as they are responsible for more than 65% of all cases of infection and lead to ~35,000 deaths in the United States annually.¹ The lack of antibiotics that can be used to treat these infections indicates the urgent need for new antimicrobial agents capable of eradicating bacterial infections.²

Venoms are an exciting new source of potential drugs and are being explored for antibiotic discovery.^3–7 Specifically, antimicrobial peptides (AMPs) have been found to be present in venoms or toxins.^3–10 AMPs are a promising alternative to conventional antibiotics as they can kill bacteria by penetrating through their membranes via non-specific membrane-related mechanisms of action. Either AMPs do not induce bacterial resistance or, if resistance occurs, it takes longer to develop than it does with conventional antibiotics.^2,11 Examples of AMPs from venoms include mastoparans EMP-EM1 (WT1, LKLMGIVKKVLGAL; Table 1) and EMP-EM2 (WT2, LKLLGIVKKVLGAI; Table 1), both of which are linear, cationic, amphipathic, and α-helical natural peptides extracted from the venom of the solitary eumenine wasp Eumenes micado (Figure 1A).¹² In addition to their antimicrobial and leishmanicidal activities, these peptides cause the degranulation of rat peritoneal mast cells.¹² However, WT1 and WT2 were also reported to have hemolytic activity for human and mouse erythrocytes at a concentration of 10⁻⁴ mol L⁻¹,¹² preventing their application to human infections. Here, WT1 and WT2 were used as templates for the structure-function-guided rational design of analogs with fine-tuned structural and physicochemical properties. Compared with their parent peptides, the new synthetic analogs were less cytotoxic for human cells and presented increased antimicrobial activity against Gram-negative and Gram-positive pathogenic bacteria both in vitro and in animals.

Table 1.

Amino acid sequence, theoretical values of normalized hydrophobicity, normalized hydrophobic moment, net charge, ratio of polar/non-polar amino acid residues, propensity to in vitro aggregation, amphiphilicity index, and values of helical fraction in water and in a mixture 3:2 of TFE and water, for WT1, WT2, and their corresponding Ala-Scan analogs

Peptides		Physicochemical properties						Helical fraction
Code	Sequence	<H>	<μH>	q	P/N	In vitro aggregation	Amphiphilicity index	fH (water)	fH (TFE:water)
WT1	LKLMGIVKKVLGAL	−0.39	0.40	4	0.56	15.42	0.79	0.05	0.34
A1–1	AKLMGIVKKVLGAL	−0.36	0.42	4	0.56	15.37	0.79	0.05	0.32
A2–1	LALMGIVKKVLGAL	−0.54	0.33	3	0.40	27.36	0.52	0.04	0.29
A3–1	LKAMGIVKKVLGAL	−0.36	0.37	4	0.56	13.64	0.79	0.07	0.29
A4–1	LKLAGIVKKVLGAL	−0.39	0.40	4	0.56	14.93	0.79	0.05	0.38
A5–1	LKLMAIVKKVLGAL	−0.40	0.39	4	0.40	29.98	0.79	0.09	0.44
A6–1	LKLMGAVKKVLGAL	−0.34	0.40	4	0.56	13.64	0.79	0.05	0.33
A7–1	LKLMGIAKKVLGAL	−0.36	0.37	4	0.56	13.65	0.79	0.09	0.31
A8–1	LKLMGIVAKVLGAL	−0.54	0.37	3	0.40	34.40	0.52	0.08	0.24
A9–1	LKLMGIVKAVLGAL	−0.54	0.29	3	0.40	56.15	0.52	0.11	0.22
A10–1	LKLMGIVKKALGAL	−0.36	0.38	4	0.56	2.75	0.79	0.09	0.23
A11–1	LKLMGIVKKVAGAL	−0.36	0.38	4	0.56	3.86	0.79	0.07	0.16
A12–1	LKLMGIVKKVLAAL	−0.40	0.39	4	0.40	102.82	0.79	0.13	0.35
A14–1	LKLMGIVKKVLGAA	−0.36	0.37	4	0.56	3.84	0.79	0.05	0.25
WT2	LKLLGIVKKVLGAI	−0.45	0.43	4	0.56	34.76	0.79	0.10	0.23
A1–2	AKLLGIVKKVLGAI	−0.41	0.45	4	0.56	34.49	0.79	0.10	0.33
A2–2	LALLGIVKKVLGAI	−0.60	0.35	3	0.40	85.50	0.52	0.15	0.29
A3–2	LKALGIVKKVLGAI	−0.41	0.41	4	0.56	27.74	0.79	0.06	0.26
A4–2	LKLAGIVKKVLGAI	−0.41	0.42	4	0.56	27.89	0.79	0.08	0.31
A5–2	LKLLAIVKKVLGAI	−0.46	0.42	4	0.40	94.76	0.79	0.10	0.30
A6–2	LKLLGAVKKVLGAI	−0.39	0.44	4	0.56	27.13	0.79	0.10	0.24
A7–2	LKLLGIAKKVLGAI	−0.41	0.40	4	0.56	27.10	0.79	0.08	0.24
A8–2	LKLLGIVAKVLGAI	−0.60	0.41	3	0.40	83.03	0.52	0.06	0.29
A9–2	LKLLGIVKAVLGAI	−0.60	0.31	3	0.40	108.91	0.52	0.06	0.31
A10–2	LKLLGIVKKALGAI	−0.41	0.41	4	0.56	10.16	0.79	0.06	0.31
A11–2	LKLLGIVKKVAGAI	−0.41	0.41	4	0.56	12.38	0.79	0.11	0.25
A12–2	LKLLGIVKKVLAAI	−0.46	0.43	4	0.40	177.88	0.79	0.08	0.38
A14–2	LKLLGIVKKVLGAA	−0.39	0.38	4	0.56	10.31	0.79	0.07	0.24

<H>, normalized hydrophobicity; <μH>, normalized hydrophobic moment; q, net charge; P/N, ratio of polar/non-polar amino acid residues; f_H, helical fraction.

Figure 1. Design, antimicrobial activity, and secondary structure elucidation of peptides from wasp venom.

(A) Schematic representation of the structure-function relationship studies, from the selection of the templates (EMP-EM1 [WT1] and EMP-EM2 [WT2]), isolated from the venom of the solitary wasp Eumenes micado, to the design of an optimized second-generation peptide.

(B) Antimicrobial activity of WT1 and WT2 and Ala-Scan analogs for the four pathogenic bacterial strains tested in this study. The red color represents bacterial growth inhibition, and the blue color represents bacterial growth.

(C and D) CD spectra of WT1 (C) and WT2 (D) and their respective Ala-Scan derivatives at 50 μmol L⁻¹ in TFE:water 3:2 v/v and water showing the conformational transition of the peptides from random coil in water to α helix in TFE:water. CD spectra were recorded in three replicates at 25°C, using a quartz cuvette with 1-mm path length, between 260 and 190 nm at 50 nm min⁻¹, with a bandwidth of 0.5 nm.

(E) Bidimensional helical wheel representations of the wild-type peptides WT1 and WT2, indicating positions where Ala-substitution decreased (blue arrows) or enhanced (red arrows) activity.

RESULTS AND DISCUSSION

Mutational scan of peptide sequences to understand structure and function

The alanine (Ala) residue is extensively used in site-directed mutagenesis to evaluate the contribution of amino acid residues to the stability, structure, and activity of proteins and peptides. This amino acid residue presents the smallest aliphatic and chemically inert side chain, which preserves the distance between neighboring residues and does not interfere with intramolecular or intermolecular interactions.³ Thus, we performed Ala screenings of both wild-type peptides (WT1 and WT2), generating 26 Ala-substituted synthetic derivatives, to assess the contribution of the side chain of each amino acid residue to the biological activities of the peptides (Figure 1A).

First, the physicochemical properties were theoretically determined. Normalized hydrophobicity, normalized hydrophobic moment, net positive charge, propensity to aggregate in vitro, and amphiphilicity index were determined by reference to the Database of Antimicrobial Activity and Structure of Peptides (DBAASP V3.0)¹³ using the Eisenberg and Weiss hydrophobicity scale for all Ala-Scan (A¹-1 to A¹⁴-1 and A¹-2 to A¹⁴-2) and WT1 and WT2 peptides (Table 1).¹⁴ The ratio of polar-to-non-polar amino acid residues and the helical wheels were determined using the HeliQuest webserver¹⁵ for all peptides (Table 1; Figure S1). The normalized hydrophobicity and hydrophobic moment of all Ala-Scan and wild-type peptides ranged from −0.34 to −0.60 and from 0.29 to 0.45, respectively (Table 1), which is within the range of other known AMPs.¹³ The hydrophobicity-related properties of peptides correlate with their antimicrobial activity as they are predictive of how peptides will interact with the lipid components of the bacterial membrane. The Ala-Scan derivatives A¹²-1, A²-2, A⁵-2, A⁸-2, A⁹-2, and A¹²-2 exhibited higher values of propensity to aggregate in vitro (102.82, 85.50, 94.76, 83.03, 108.91, and 177.88, respectively) compared with WT1 and WT2 (15.42 and 34.76, respectively; Table 1). High values of propensity to aggregate in vitro are usually associated with decreased physical stability and antimicrobial activity, as well as increased cytotoxicity.¹⁶

In addition, most Ala-Scan derivatives presented an amphiphilicity index of 0.79 (Table 1). Because the values for the amphiphilicity index were higher than zero, all peptides showed helical structural stability at the membrane-water interface.¹⁷ Next, we tested WT1, WT2, and their Ala-Scan derivatives for activity against the Gram-negative bacteria Escherichia coli ATCC 11775 and Pseudomonas aeruginosa strains PAO1 and PA14, as well as the Gram-positive bacterial strain Staphylococcus aureus ATCC 12600, based on their minimal inhibitory concentration (MIC) values (Figures 1B and S2). We performed minimal bactericidal concentration (MBC) experiments, which showed that the peptides killed the bacterial cells at the observed MIC conditions. Thus, we confirmed that the MIC and MBC values were the same by counting the colony-forming units (CFUs) after 21 h of incubation at 37°C.

WT1 and WT2 sequences are amphipathic, and, similarly to other mastoparans, they are α-helical in the presence of helix-inducing solvents.¹² To analyze the contribution of each residue’s side chain to the helical structure, we performed circular dichroism (CD) experiments. In each case, the helical fraction (f_H) was determined from the negative band at 222 nm in a mixture of 2,2,2-trifluoroethanol (TFE) and water (TFE: water, 3:2, v/v) and in water alone (Table 1, and Figures 1C and 1D). We chose a TFE:water (3:2, v/v) solution because it is extensively used to study peptide secondary structure; this solution induces helical conformation in peptides by promoting dehydration and, consequently, intramolecular hydrogen bond formation.^18,19 All Ala-Scan derivatives were unstructured in water (0.04 ≤ f_H ≤ 0.15) but presented a helical structure in TFE:water (0.22 ≤ f_H ≤ 0.44), except peptide A¹¹-1, which was unstructured in both water and TFE:water. This common feature of small amphipathic cationic AMPs, known as the helical-coil transition,²⁰ occurs at hydrophilic-hydrophobic interfaces (Table 1; Figures 1C and 1D).

WT1 presented antimicrobial activity with MICs of 8 μmol L⁻¹ for E. coli ATCC 11775 and 16 μmol L⁻¹ for S. aureus ATCC 12600, and MICs of 32 μmol L⁻¹ for P. aeruginosa PAO1 and PA14 (Figures 1B and S2). WT2 was active with MICs of 16 μmol L⁻¹ for E. coli ATCC 11775, 32 μmol L⁻¹ for S. aureus ATCC 12600 and P. aeruginosa PAO1, and 64 μmol L⁻¹ for P. aeruginosa PA14 (Figures 1B and S2). Ala-Scan screening studies revealed that, when glycine (Gly) residues were replaced by Ala in the hydrophilic face of WT1 and WT2 sequences at positions 5 and 12 (Figure 1E), the predicted normalized hydrophobicity and observed helicity of the modified peptides increased compared with those of the wild-type peptides (Table 1; Figures 1C and 1D). The Gly residue presents the less hindered side chain (one hydrogen atom), thus leading to a high degree of freedom for the peptide, allowing different conformational intermediates, including helical steps.

The antimicrobial activities of A⁵-1, A¹²-1, A⁵-2, and A¹²-2 were significantly increased against E. coli ATCC 11775 and S. aureus ATCC 12600 (Figures 1B, S2A, and S2B). Synthetic peptide A⁵-1 was up to 4-fold more active against E. coli ATCC 11775 and S. aureus ATCC 12600 than WT1, with an MIC of 4 μmol L⁻¹ for both bacterial strains. Peptide A⁵-2 was 4-fold more active against E. coli ATCC 11775 and 8-fold more active against S. aureus ATCC 12600 compared with WT2, with MICs of 4 μmol L⁻¹ for both bacterial strains (Figures 1B, S2A, and S2B). A¹²-1 presented an MIC of 2 μmol L⁻¹ for E. coli ATCC 11775 and S. aureus ATCC 12600, 4- and 8-fold higher than WT1, respectively. A¹²-2 was 4-fold more active against E. coli ATCC 11775 and 16-fold more active against S. aureus ATCC 12600 than WT2, with MICs of 4 and 2 μmol L⁻¹, respectively (Figures 1B, S2A, and S2B).

When the Met residue at position 4 of the WT1 was substituted by Ala (Figure 1E), the resulting peptide, A⁴-1, presented a higher helical fraction value and increased antimicrobial activity against E. coli ATCC 11775 (MIC = 4 μmol L⁻¹), and S. aureus ATCC 12600 (MIC = 8 μmol L⁻¹) compared with WT1 (Table 1; Figures 1B, 1C, S2A, and S2B). The α helix structure of A⁴-1 was stabilized because Ala has the highest helical propensity (0 kcal mol⁻¹) compared with all other amino acids, whereas Met has a lower helical propensity (0.24 kcal mol⁻¹) compared with Ala.²¹ Replacement of Leu at the interface between the hydrophilic and hydrophobic faces (position 4; Figure 1E) and one helical step from the charged residue lysine (Lys) (position 8; Figure 1E) of WT2 by Ala stabilized the hydrophobic face, increased the helicity, and improved the antimicrobial activity of A⁴-2 against E. coli ATCC 11775 and S. aureus ATCC 12600 (Table 1; Figures 1B, 1D, S2A, and S2B). The insertion of Ala within the hydrophilic face of WT1 and WT2 (positions 4, 5, and 12, red arrows; Figure 1E) increased antimicrobial activity and stabilized the α-helical structure. Conversely, structural modifications of the hydrophobic face of WT1 and WT2 (positions 3, 6, 11, and 14, blue arrows, Figure 1E) decreased antimicrobial activity (Figures 1B and S2).

Torres et al. obtained similar results with an Ala-Scan screening of the mastoparan peptide polybia-CP, isolated from the venom of a tropical wasp species. Polybia-CP showed enhanced antibacterial activity upon Ala substitutions in the hydrophilic face or at the interface between the hydrophobic and hydrophilic faces of the peptide.³ In addition, modifications at the hydrophobic face of polybia-CP also decreased antimicrobial activity, as observed for WT1 and WT2.³ The presence of the hydrophobic amino acid Ile at the interface between the hydrophilic and hydrophobic faces of WT1 and WT2 (position 6, Figure 1E) was important for their antimicrobial activity since A⁶-1 and A⁶-2 presented lower activity (higher MIC values) against S. aureus ATCC 12600 and P. aeruginosa PAO1 and PA14 compared with both wild-type peptides (Figures 1B and S2B–S2D). Ala substitutions of Leu at positions 3, 11, and 14 of WT1, and positions 3 and 11 of WT2, also resulted in a loss of activity (Figures 1B, 1E, and S2), showing that hydrophobicity was important for the antimicrobial activity of these peptides. A³-1, A¹¹-1, and A¹⁴-1 presented lower values of helical fraction compared with WT1, suggesting a relationship between helicity and activity; however, this relationship was not observed for A³-2 or A¹¹-2 (Table 1; Figures 1B–1D). Moreover, the residues at the end of the sequence, Leu for WT1 and Ile for WT2, were important for the antimicrobial activity of these natural peptides (Figures 1B, 1D, and S2).

Collectively, the results obtained by the Ala-Scan screening revealed a correlation between helicity and antimicrobial activity, especially for WT1-derived Ala-Scan peptides. The exception was A¹-1, as the replacement of Leu by Ala in the hydrophilic face of WT1 (position 1, Figure 1E) resulted in increased activity against all bacteria tested (MIC = 4 μmol L⁻¹ for E. coli ATCC 11775, 8 μmol L⁻¹ for S. aureus ATCC 12600, and 16 μmol L⁻¹ for P. aeruginosa PAO1 and PA14; Figure S2) without increasing the helical fraction value of A¹-1 in relation to WT1 (Table 1; Figure 1C). For the WT2 derivatives, the correlation between helicity and activity was not clear, since all WT2-derived Ala-Scan peptides had higher helical fraction values compared with WT2 (Table 1). To further investigate the physicochemical and structural aspects influencing the antimicrobial activity of these natural peptides, we decided to design new derivatives of WT1 and WT2 (Table 2).

Table 2.

Properties of peptides WT1, WT2, and their corresponding second-generation analogs

Peptides		Physicochemical properties						Helical fraction
Code	Sequence	<H>	<μH>	q	P/N	In vitro aggregation	Amphiphilicity index	fH (water)	fH (TFE:water)
WT1	LKLMGIVKKVLGAL	−0.39	0.40	4	0.56	15.42	0.79	0.05	0.34
I14–1	LKLMGIVKKVLGAI	−0.42	0.42	4	0.56	28.33	0.79	0.19	0.41
L4–1	LKLLGIVKKVLGAL	−0.42	0.41	4	0.56	21.87	0.79	0.14	0.88
K12–1	LKLMGIVKKVLKAL	−0.25	0.53	5	0.56	1.76	1.05	0.06	0.62
K13–1	LKLMGIVKKVLGKL	−0.24	0.47	5	0.75	1.77	1.05	0.14	0.13
WT2	LKLLGIVKKVLGAI	−0.45	0.43	4	0.56	34.76	0.79	0.10	0.23
K1–2	KKLLGIVKKVLGAI	−0.26	0.57	5	0.75	31.51	1.05	0.12	0.62
K6–2	LKLLGKVKKVLGAI	−0.24	0.48	5	0.75	26.61	1.05	0.10	0.41
K10–2	LKLLGIVKKKLGAI	−0.26	0.35	5	0.75	8.22	1.05	0.14	0.50
K13–2	LKLLGIVKKVLGKI	−0.29	0.51	5	0.75	8.15	1.05	0.15	0.89

Theoretical values of: normalized hydrophobicity (<H>); normalized hydrophobic moment (<μH>); net charge (q); ratio of polar/non-polar amino acid residues (P/N); propensity to aggregate in vitro; amphiphilicity index; and values of helical fraction (f_H) in water and in a mixture 3:2 of TFE and water.

Structure-function-guided design of a new generation of synthetic peptides

Since the Ala-substituted analogs were not significantly more active than the templates, we decided to go back to the templates and assess how to tune them in order to increase their antimicrobial activity while preserving their tendency to structure helically. To achieve this, we designed a second generation of WT1 and WT2 peptides with single Lys substitutions, yielding six synthetic derivatives. Lys was selected given its tendency to lead to increased antimicrobial activity while not increasing cytotoxicity in sequences derived from venoms.^{3,4,6,7,22–25} Lys residues were used instead of arginine (Arg) because Arg is more likely to increase the toxicity of short amphipathic peptides such as the EMP-EM templates.²⁶ We placed the single substitutions in positions that were expected to lead to analogs with higher amphipathicity and lower in vitro aggregation tendencies (Table 2), properties that proved to be important for the antimicrobial activity of this peptide family. To explore the two only distinguishing amino acid residues between WT1 (Met and Leu) and WT2 (Leu and Ile) (positions 4 and 14), we synthesized derivatives L⁴-1 and I¹⁴-1.

First, the physicochemical properties (Table 2) and helical wheel projections (Figure S3) of all new derivatives were predicted using DBAASPDBAASP⁸⁰ and HeliQuest, respectively.^13–15 The second-generation derivatives K¹²-1, K¹³-1, K¹⁰-2, and K¹³-2 presented lower values of propensity to aggregate in vitro (1.76, 1.77, 8.22, and 8.15, respectively, Table 2) compared with WT1, WT2, or Ala-Scan analogs (Table 1). These low values of propensity to aggregate in vitro are related to the hydrophobic balance of the peptide, which directly affects its propensity to interact with lipid membranes and, consequently, its antimicrobial activity and cytotoxicity.¹⁶ Moreover, the second-generation derivatives showed a higher value of amphiphilicity index of 1.05 in relation to derivatives WT1, WT2, and Ala-Scan derivatives (except I¹⁴-1 and L⁴-1), indicating a gain in helical structural stability at the membrane-water interface (Tables 1 and 2). All eight derivatives from the second generation were synthesized and tested against the pathogenic bacteria E. coli ATCC 11775, S. aureus ATCC 12600, and P. aeruginosa PAO1 and PA14 (Figures 2A and S4). CFU counts of bacteria after 21 h of incubation at 37°C were used to determine the minimum bactericidal concentrations (MBCs). The secondary structures of this second generation of peptides were analyzed by CD spectroscopy measurements in TFE:water (3:2 v/v) and water (Figure 2B), and the helical fraction values were also calculated (Table 2). All derivatives were random coils in water and presented a higher f_H value in TFE:water compared with WT1 and WT2, except for K¹³-1, which adopted a random coil in water and in TFE:water (Figure 2B), showing that the Ala residue in position 13 was important for the helical structure of WT1. On the contrary, this random-coil structure in water and in TFE:water was not observed for WT2 when we substituted the Ala residue in position 13 by a Lys residue (K¹³-2).

Figure 2. Antimicrobial activity and elucidation of secondary structure.

(A) Antimicrobial activity of WT1, WT2, and second-generation analogs for all tested pathogenic bacteria. The red color represents bacterial growth inhibition, and the blue color represents bacterial growth. Heat maps obtained directly from OD measurements of 96-well plates after treatment are shown in Figure S4.

(B) CD spectra of WT1 and WT2 and their respective second-generation derivatives at 50 μmol L⁻¹ in TFE:water 3:2 v/v, showing α helix conformation, and in water, showing random-coil conformation. CD spectra were recorded in three replicates at 25°C, using a quartz cuvette with 1-mm path length, between 260 and 190 nm at 50 nm min⁻¹, with a bandwidth of 0.5 nm.

Exploration of derivatives I¹⁴-1 and L⁴-1 showed that I¹⁴-1 had a higher helical fraction value than WT1, but it was more active than WT1 only against P. aeruginosa PAO1 (Table 2, Figures 2A, 2B, and S4A). L⁴-1 presented a lower predicted normalized hydrophobicity and higher values of helical fraction (0.88) and normalized hydrophobic moment compared with A⁴-1 or WT1 (Tables 1 and 2). Despite its higher helicity, L⁴-1 showed no significant improvement in activity compared with A⁴-1. As previously discussed, when Gly at position 12 of WT1 was replaced by Ala, the resulting analog, A¹²-1, was more active than WT1 against E. coli ATCC 11775 and S. aureus ATCC 12600 (MIC = 2 μmol L⁻¹) but showed no significant improvement in activity against P. aeruginosa strains PAO1 and PA14 compared with WT1 (Figure 1B and S2).

To further explore the effectiveness of these peptides, particularly against Gram-negative bacteria, a Lys residue was used to increase the positive net charge and decrease the flexibility and hydrophobicity of the WT1 and WT2 sequences. Gly was replaced by Lys at position 12 of the WT1 sequence, leading to derivative K¹²-1, which, compared with WT1, had a higher helical fraction in TFE:water (0.62, Table 2) and was 8-fold more active against P. aeruginosa PAO1, 4-fold more active against S. aureus ATCC 12600 and P. aeruginosa PA14, and 2-fold more active against E. coli ATCC 11775 (Figures 2A and S4). Lys-for-Gly substitutions in mastoparans usually increase antimicrobial activity, mainly because this family of peptides is known for their secondary-structure-dependent antimicrobial activity and Lys is a positively charged residue that stabilizes the secondary structure.³ On the contrary, Gly is highly flexible, favoring intermolecular interactions and destabilizing the secondary structure.³ Interestingly, analog K¹³-1, which presented a random-coil secondary structure in water and TFE:water (Figure 2B), was 4-fold more active against P. aeruginosa PAO1 (MIC of 8 μmol L⁻¹) and 2-fold more active against E. coli ATCC 11775 (MIC of 4 μmol L⁻¹) and P. aeruginosa PA14 (MIC of 16 μmol L⁻¹) than WT1 (Figures 2A and S4). A positive charge was added to this derivative at the interface between the hydrophilic and hydrophobic faces when the Ala residue, at position 13 of the WT1 sequence, was replaced by Lys (Figure S3). On the other hand, such modification at the interface caused an 8-fold decrease in the activity of K¹³-1 against S. aureus ATCC 12600, increasing the MIC to 128 μmol L⁻¹ (Figures 2A and S4B).

To assess the effect of a positive charge at the N-terminal extremity of the WT2 sequence, a Leu residue was substituted by Lys, leading to the synthetic derivative K¹-2 (Table 2). Such replacement at the N-terminal extremity stabilized the α helix structure, and K¹-2 was up to 4-fold more active against Gram-negative bacteria than WT2 (Table 2; Figures 2A and S4). In addition, Ile and Ala residues were substituted by Lys at the interface between the hydrophilic and hydrophobic faces of WT2, yielding derivatives K⁶-2 and K¹³-2, respectively. K⁶-2 was 2-fold less active against E. coli ATCC 11775 than WT2 and was not active against S. aureus ATCC 12600 at any of the concentrations tested (Figures 2B and S4). These results confirm the importance for its antimicrobial activity of the Ile residue at these positions within the WT2 sequence, as previously observed for A⁶-2, since this derivative also lost activity when Ile was replaced by Ala (Figures 1B, 1E, and S2). K¹³-2 exhibited 2-fold decreased activity against S. aureus ATCC 12600; however, this peptide presented up to 8-fold greater activity against all tested Gram-negative bacteria compared with WT2, with MICs of 4 μmol L⁻¹ for E. coli ATCC 11775 and P. aeruginosa PAO1, and 16 μmol L⁻¹ for P. aeruginosa PA14, which may derive from its high helical fraction (0.89, Table 2; Figures 2A, 2B, and S4). Finally, to test the effect of inserting a positive charge on the hydrophobic face of WT2, Val was replaced by Lys, yielding K¹⁰-2. Peptide K¹⁰-2 showed greater helicity; 4-fold higher antimicrobial activity against E. coli ATCC 11775, and S. aureus ATCC 12600; and 8-fold higher antimicrobial activity against P. aeruginosa PAO1 and PA14 compared with WT2, with MICs of 4 μmol L⁻¹ for E. coli ATCC 11775 and P. aeruginosa PAO1, and 8 μmol L⁻¹ for S. aureus ATCC 12600 and P. aeruginosa PA14 (Table 2; Figures 2A, 2B, and S4). Thus, replacing residues from the original sequence with a Lys residue at strategic positions of the WT1 and WT2 sequences, which increased the net positive charge of these peptides, significantly improved their antimicrobial activity against Gram-negative bacteria (except for K⁶-2), although it decreased the activity of K¹³-1, K⁶-2, and K¹³-2 against S. aureus ATCC 12600 (Figures 2A and S4). Nonetheless, K¹²-1 and K¹⁰-2 were more active against all tested bacteria compared with their corresponding templates, WT1 and WT2 (Figures 2A and S4).

Mechanism of action, synergy, and evolution of bacterial resistance

Since AMPs, such as mastoparans, generally establish the first contact with the bacterial cell membrane by electrostatic and hydrophobic interactions, these peptides can act through different mechanisms of action (MoAs) and any structural modification can affect their MoAs.^8,27,28 For MoA studies, we selected P. aeruginosa PAO1 as a model bacterial strain, because of its medical importance as a multidrug-resistant pathogen,²⁹ and the following peptides: (1) both templates WT1 and WT2; (2) K¹²-1 for being the most active peptide from WT1 family against Gram-negative and Gram-positive bacterial strains (Figures 2A and S4); (3) K¹³-1 for its activity against the Gram-negative strains tested and its random-coil secondary structure (Table 2; Figure 2B); (4) K¹³-2 also for its activity against the Gram-negative strains tested and for presenting the highest value of helical fraction among all studied peptides (Table 2); and (5) polymyxin B (PMB) for being a well-known membrane-disrupting peptide.³⁰

To assess whether these peptides permeabilize bacterial outer membranes, we used the fluorescent probe (1-(N-phenylamino)-naphthalene (NPN). In aqueous environments, NPN emits weak fluorescence and can only permeate bacterial outer membranes when those membranes are damaged.³¹ When this probe interacts with the lipidic environment of damaged outer membranes, it emits fluorescence at increased intensity, indicating that the membrane has been permeabilized by the peptide (Figure 3A).³¹ The NPN assay showed that all tested peptides permeated the outer membrane of P. aeruginosa PAO1, as observed for PMB (Figure 3B). To evaluate whether these peptides depolarize the cytoplasmic membrane of P. aeruginosa PAO1, we utilized the probe 3,3′-dipropylthiadicarbocyanine iodide (DiSC₃-5). DiSC₃-5 is a potentiometric probe that accumulates in cytoplasmic membranes and aggregates at high concentrations, causing fluorescence quenching. When the cytoplasmic membrane is destabilized, DiSC₃-5 migrates to the cytoplasm or to the external environment, emitting increased fluorescence intensity (Figure 3C).³¹ The DiSC₃-5 assay revealed that all tested peptides depolarized the cytoplasmic membrane of P. aeruginosa PAO1 more efficiently than PMB (Figure 3D). K¹²-1 was the most efficient peptide in destabilizing the outer and cytoplasmic membranes of P. aeruginosa PAO1 compared with other peptides, corresponding to its high antimicrobial activity against all tested bacteria (Figures 2A, 3B, and 3D). The random-coil peptide K¹³-1, such as all the tested α-helical peptides, also permeated the outer membrane and depolarized the cytoplasmic membrane of P. aeruginosa PAO1 (Figures 3B and 3D). This was an interesting observation that revealed that the mechanism of action of these peptides was not entirely driven by their secondary structure.

Figure 3. Mechanism of action, synergy, and bacterial resistance.

(A) Schematic representation of the NPN assay, in which molecules of NPN (represented by gray spheres) present weak fluorescence emission intensity in an aqueous environment. When the outer membranes are permeabilized by peptides, the NPN molecules interact with the lipidic environment of damaged outer membranes and the intensity of blue fluorescence emission increases (represented by blue spheres).

(B) NPN graph for outer membrane permeabilization of Pseudomonas aeruginosa PAO1 by polymyxin B (PMB), WT1, K¹²-1, K¹³-1, WT2, and K¹³-2 peptides. Profiles with a rapid increase in fluorescence emission intensity, followed by a slow decay, were obtained after measurement of white 96-well plates on a Thermo Scientific Varioskan LUX fluorescence spectrophotometer, with the excitation wavelength set to 350 nm and the emission wavelength set to 420 nm, according to the experimental procedure described in the section “experimental procedures.” All NPN assays were done in three replicates, including the controls, which consisted of only HEPES solution, HEPES solution and NPN (not shown), HEPES solution and P. aeruginosa PAO1 (not shown), and HEPES solution with both P. aeruginosa PAO1 and NPN. Data are represented as mean ± SD.

(C) Schematic representation of the DiSC₃-5 assay, in which molecules of DiSC₃-5 (represented by gray spheres) accumulate in cytoplasmic membranes and aggregate at high concentrations, causing fluorescence quenching. When the cytoplasmic membrane is destabilized by peptides, DiSC₃-5 migrates to the cytoplasm or to the external environment, and red fluorescence emission intensity (represented by red spheres) increases.

(D) DiSC₃-5 graph for cytoplasmic membrane depolarization of P. aeruginosa PAO1 by PMB, WT1, K¹²-1, K¹³-1, WT2, and K¹³-2 peptides. Profiles with increases and decreases in fluorescence emission intensity were obtained after measurement of black 96-well plates on a Thermo Scientific Varioskan LUX fluorescence spectrophotometer, with the excitation wavelength set to 622 nm and emission wavelength set to 670 nm as described in the section “experimental procedures.” DiSC₃-5 graph obtained after the addition of triton solution is shown in Figure S5. All DiSC₃-5 assays were done in three replicates, including the controls, which consisted of only HEPES solution, and HEPES solution containing PAO1 and DiSC₃-5. Data are represented as mean ± SD.

(E) Synergy assay for activity against P. aeruginosa PAO1 between ciprofloxacin, ofloxacin, gentamicin, polymyxin B, or erythromycin, and each of four peptides: WT1, WT2, K¹²-1, and K¹³-1. The Fractional Inhibitory Concentration Index (FICI) values, which indicate the degree of synergy between two antimicrobial agents against a target microorganism, were calculated based on the MICs of WT1, WT2, K¹²-1, and K¹³-1 and the commercial antibiotics used alone and in combination. FICI values <0.5 indicate synergy; 0.5 < FICI < 1 indicates additive effects; 1 < FICI < 4 indicates indifference; and FICI > 4 indicates antagonism (not represented in the graph).

(F) Resistance assay: development of resistance to ciprofloxacin, PMB, WT1, K¹²-1, and K¹³-1 in Escherichia coli ΔmutS. The experiment was performed for 20 days as described in detail in the section “experimental procedures.” Data are represented as mean ± SD.

On the other hand, K¹³-2, the peptide with the highest helical fraction value among all second-generation derivatives (Table 2), was the least efficient in destabilizing the outer and cytoplasmic membranes, further demonstrating that the MoAs of these peptides do not depend entirely on their secondary structure (Figures 3B and 3D).

To evaluate whether WT1, WT2, K¹²-1, and K¹³-1 can interact synergistically with antibiotics with different MoAs, a synergy assay was performed (Figure 3E). First, seven antibiotics (i.e., ciprofloxacin, metronidazole, ofloxacin, gentamicin, PMB, erythromycin, and chloramphenicol) were tested for activity against P. aeruginosa PAO1 (Figure S6). As metronidazole and chloramphenicol were not active against P. aeruginosa PAO1 at the highest concentration tested (128 μmol L⁻¹; Figure S6), the other five of these antibiotics were selected for further testing. These five antibiotics were diluted using the microdilution technique at concentrations ranging from 2- to 0.03-fold MIC and combined with WT1, WT2, K¹²-1, and K¹³-1 with the same range of concentrations (Figure S7). No synergistic effect was observed when these peptides were combined with any of the five commercial antibiotics (Figures 3E and S7). Clarifying the synergistic interactions (or lack thereof) of antibiotic combinations requires a deep understanding about the MoAs underlying the biological activity of each drug. Our MoA studies revealed that these peptides can rapidly kill Gram-negative bacteria such as P. aeruginosa PAO1 by efficiently permeabilizing their outer and cytoplasmic membranes (Figures 3B and 3D). However, additional intra-cellular modes of action involved in the antimicrobial activity of these peptides have not yet been characterized and warrant further studies. Based on the mechanistic results obtained here, we propose that these peptides can effectively kill Gram-negative bacteria at the membrane level. However, these membrane-targeting mechanisms of the peptides do not necessarily lead to synergy with conventional antibiotics. A deeper understanding of the specific modes of action involved in the antimicrobial activity of these peptides is needed.

Although numerous papers have been published describing the synergistic effects between membrane-permeating peptides and antibiotics whose targets are intracellular, this tendency is not necessarily widespread.^32–38 For example, studies have addressed the lack of synergy between membrane-disrupting cationic AMPs and conventional antibiotics.^{32–34,39,40} Giacometti et al. reported that magainin II synergized with β-lactam antibiotics, but no synergy was observed when magainin II was combined with other classical antibiotics.^32–34 On the other hand, Ulvatne et al. reported a lack of synergy when certain synthetic peptides were combined with β-lactam antibiotics. The authors suggested that their peptides interacted with bacterial membranes by alternative modes of action independently of pore formation.⁴⁰

The lowest fractional inhibitory concentration indices (FICIs) were obtained when PMB was combined with WT1 (0.7), WT2 (0.5), K¹²-1 (0.7), or K¹³-1 (0.6) (Figures 3E and S7), indicating an additive effect for each of these peptides combined with PMB against P. aeruginosa PAO1. An additive effect is equivalent to the sum of the effects of the individual components; thus, these peptides may act through the same MoAs as PMB. PMB acts via a self-promoted uptake pathway by binding to negatively charged phosphate groups on the lipopolysaccharide in the outer membrane of Gram-negative bacteria.³⁰ PMB destabilizes the outer membrane and permeabilizes it and can penetrate the cytoplasmic membrane by binding to phospholipids, causing lethal leakage of cytoplasmic components.³⁰ An additive effect for the combination of WT1, WT2, K¹²-1, or K¹³-1 and PMB is in agreement with the results of our MoA studies. Thus, the activity of these peptides does not synergize with that of PMB, since these peptides can permeate the outer membrane and depolarize the cytoplasmic membrane more efficiently than PMB (Figures 3B and 3D). As both templates are produced within the same venom, we hypothesized that WT1 and WT2 act synergistically in the venom mixture. Therefore, WT1 and WT2 were tested in combination at different concentrations against the pathogenic strain P. aeruginosa PAO1. We observed an additive effect (FICI = 0.7), as previously obtained by the combinations of the template peptides and PMB (Figure S7).

To evaluate whether WT1, K¹²-1, and K¹³-1 select for pathogen resistance, a resistance development assay was performed with E. coli JW2703 (hypermutant strain) ΔmutS::kan. (from the Keio collection), hereafter referred to as E. coli ΔmutS for simplicity; ciprofloxacin, a broad-spectrum second-generation fluoroquinolone that acts by inhibiting DNA replication; and PMB, a known membrane disrupting agent.⁴¹ Ciprofloxacin was used as a positive control because it is well known to rapidly trigger resistance development in bacteria.^42–45 The strain E. coli JW2703 ΔmutS::kan was selected for the resistance development experiment because it is more relevant than wild-type E. coli or P. aeruginosa strains due to its ability to rapidly mutate, making it an excellent strain for resistance development assays as previously reported.⁴⁶ This strain is mutated in its mutS gene, yielding a hypermutant strain. The development of resistance in E. coli ΔmutS was assessed by monitoring changes in the MICs of WT1, K¹²-1, K¹³-1, and ciprofloxacin induced by increasing concentrations of treatment over 20 days (Figure 3F). Treatment with the fluoroquinolone ciprofloxacin induced bacterial resistance within 4 days from the onset of the experiment, and the MIC of ciprofloxacin increased by 1,000-fold after 18 days (Figure 3F). Conversely, the MICs of WT1, K¹²-1, K¹³-1, and PMB did not vary significantly over the same period. Thus, these peptides killed E. coli ΔmutS without selecting for peptide-resistant mutants (Figure 3F). Bacterial killing likely occurs through non-specific MoAs; once such peptides disrupt the outer and cytoplasmic membrane of bacteria, the cytoplasmic content leaks out, causing bacterial cell death.⁴⁷

Cytotoxicity profile studies and in vivo antimicrobial activity in a mouse model

P. aeruginosa is a pathogenic bacterium that causes pneumonia and infections of the skin, the urinary tract, and the gastrointestinal tissue.^48–51 This bacterium has developed resistance to available antibiotics,⁵² and antimicrobial peptides have been considered potential candidates for the treatment of these infections.⁵³ First, peptides WT1, WT2, one Ala-substituted analog, and the second-generation derivatives that presented potent activity against P. aeruginosa PAO1 (i.e., A¹⁴-1, K¹²-1, K¹³-1, K¹⁰-2, and K¹³-2) were tested for cytotoxicity against human embryonic kidney (HEK) cells (ATCC HEK293T; Figure 4A) and primary human keratinocytes (Figure 4B). Primary human keratinocytes were used as a proxy for analyzing the toxicity of peptides in contact with the skin, and HEK cells were used in cytotoxicity assays because many antibiotics are excreted by the kidney and a requirement for promising new drugs is that they be non-nephrotoxic.⁵⁴ Peptide concentrations used in the cytotoxicity analysis ranged from 1 to 64 μmol L⁻¹, as the MICs of all tested peptides were below 64 μmol L⁻¹ against all tested bacteria. When treated with the peptides WT1, WT2, and K¹²-1 at their corresponding MIC for P. aeruginosa PAO1, the keratinocytes retained 70%, 86%, and 96% cell viability, respectively (Figures 4B and S4C). WT1 at its MIC showed toxicity for HEK cells, with cell viability of 75%, while WT2 and K¹²-1 were toxic for HEK cells at 2- and 4-fold MIC, with cell viability of 13% and 65%, respectively (Figures 4A and S4C). K¹⁰-2 exhibited no toxicity against keratinocytes at 8 μmol L⁻¹ (with 92% of cell viability remaining) and against HEK cells at 16 μmol L⁻¹ (i.e., 2- and 4-fold higher than its MIC against P. aeruginosa PAO1; Figures 4A, 4B, and S4C). The most promising results were obtained for K¹³-1 and K¹³-2 (Figures 4A and 4B). These peptides presented a low toxicity against keratinocytes at the highest concentration tested (i.e., 64 μmol L⁻¹, from 8- to 16-fold higher than their MIC, against P. aeruginosa PAO1; Figures 4B and S4C), thus proving a pronounced therapeutic window. K¹³-1 was toxic for HEK cells at 4-fold MIC, and A¹⁴-1 and K¹³-2 presented no toxicity against HEK cells at any of the concentrations tested (Figures 4A and S4C). The peptides K¹³-1 and K¹³-2 are more cationic and amphipathic than their templates (WT1 and WT2, respectively) and the first-generation peptides. Thus, it is to be expected that there will be a weaker interaction between such peptides and eukaryotic membranes compared with bacterial membranes, since eukaryotic membranes are more hydrophobic and only slightly negatively charged.^55,56

Figure 4. Cytotoxicity and in vivo studies.

(A and B) Cytotoxic activity of WT1, A¹⁴-1, K¹²-1, K¹³-1, WT2, K¹⁰-2, and K¹³-2 against (A) human embryonic kidney cells (HEK293T) and (B) primary human keratinocytes.

(C) Schematic representation of the in vivo assay procedure. The mice were anesthetized with isoflurane and weighed; their backs were shaved, and a superficial linear skin abrasion was made using a needle to damage the stratum corneum and upper layer of the epidermis. Then 50 μL of 10⁷ CFU mL⁻¹ in phosphate-buffered saline (PBS) of P. aeruginosa PAO1 was inoculated over the scratch in the back of the mice. After 1 h, peptide solutions in PBS at 32 μmol L⁻¹ for K¹³-2 and 16 μmol L⁻¹ for K¹²-1 and K¹³-1 were added to the infected area. This procedure was done for four mice per peptide tested. After 2 days, mice from each group were killed and weighed, and the area of scarified skin was cut, homogenized using a bead beater for 20 min (25 Hz), and serially diluted for CFU quantification. This procedure was repeated after 4 days with the mice from each group. Two technical replicates were performed for each sample to ensure accuracy.

(D) Mice weight monitoring for potential in vivo toxicity assessment. The body weight of infected mice was normalized by the body weight of uninfected mice. Data are represented as mean ± SD.

(E) Anti-infective activity of K¹²-1, K¹³-1, and K¹³-2 in vivo compared with control groups. Statistical significance was determined using one-way ANOVA followed by Dunnett’s test; p values are shown in the graph.

Therefore, based on their optimal toxicity profiles, peptides K¹²-1, K¹³-1, and K¹³-2 were selected for in vivo studies at the safe concentrations of 16, 16, and 32 μmol L⁻¹, respectively, in a skin scarification mouse model (Figure 4C).^6,57–59 To assess peptide toxicity in vivo, mice were weighed before and after treatment to monitor weight variation, since variations of up to 20% are a widely used proxy of distress, morbidity, and overall toxicity.^31,60,61 We also monitored the mice for toxicity markers such as itchiness,^62,63 redness,⁶⁴ and swelling.^65,66 None of the tested peptides showed side effects or toxicity in vivo, indicated by no significant change in the body weight of the treated mice compared with untreated mice (Figure 4D). Skin infection was induced by administering a P. aeruginosa PAO1 solution at 10⁷ CFU mL⁻¹ on the back of mice previously scratched with a needle and treated witha single dose of peptide solutions (Figure 4C). After 2 days of treatment, K¹²-1, K¹³-1, and K¹³-2 reduced bacterial quantities by 38-, 112-, and 4,600-fold, respectively, compared with the untreated control group of mice (Figure 4E). The peptide K¹³-2 showed potent bactericidal activity in vivo after 2 days of treatment. After 4 days of treatment, K¹²-1 still inhibited the proliferation of bacterial cells (i.e., showed bacteriostatic activity), while K¹³-1 had a bactericidal effect (Figure 4E). Peptides K¹³-1 and K¹³-2 did not completely sterilize the infection but reduced bacterial loads by 1,050- and 13,460-fold (three and five orders of magnitude) compared with the untreated control group, respectively, with single-dose administration at low concentrations (16 and 32 μmol L⁻¹, respectively) after 4 days of treatment (Figure 4E).

To conclude, drug-resistant bacterial infections have become a serious public health problem that needs to be addressed with novel strategies. AMPs are potential candidates for new antibiotics because their use is unlikely to select for resistance due to the use of simultaneous and diverse MoAs toward different targets in bacteria.^67,68 Venoms, which serve the producing organism as a means of defense and predation, are a source of a multitude of AMPs.^4,68,69 Here, we used a structure-function-guided design approach to fine-tune the physicochemical features of EMP-EM1 (WT1) and EMP-EM2 (WT2), two mastoparan-like peptides isolated from E. micado wasp venom. First, we unraveled the importance of each amino acid residue to their antimicrobial activity by performing an Ala-Scan screening. We then evaluated the role of their secondary structure (α helix), net charge, and the positions of certain residues within their hydrophilic face and interface in relation to antimicrobial activity. Altogether, Ala substitutions within the hydrophobic face at positions 3, 11, and 14, and at position 6, on the interface, significantly decreased the antimicrobial activity of the peptides (Figures 1B and 1E). Conversely, Ala substitutions at positions 5 and 12 within the hydrophilic face, and at position 4 within the interface, increased the antimicrobial activity and helical fraction. When a Gly residue was replaced by Ala at positions 5 and 12 within the WT1 and WT2 sequences (Figure 1E), the degree of freedom for these peptide structures decreased, favoring the stabilization of the helical structure, which proved to be crucial for the antimicrobial activity of this class of peptides.³ For peptide A⁴-1, as the Met residue has a lower helical propensity value than Ala,²¹ replacing Met by Ala can influence the stability of this peptide’s secondary structure.

The rational design of a second generation of synthetic peptides based on Lys substitutions resulted in peptides with lower theoretical values for in vitro aggregation and increased amphipathicity, which were also less toxic in vitro (Figures 4A and 4B) and possessed increased antimicrobial activity against Gram-negative bacteria (Figure 2A), compared with their respective templates WT1 and WT2 (Figures 1B, 4A, and 4B). Conversely, peptides with positive net charge values higher than the template showed a decrease in antimicrobial activity against the Gram-positive bacterium S. aureus. This finding could help to lay the foundation for the future development of narrow-spectrum peptides aimed at treating Gram-negative bacteria. These Lys-substituted peptides killed bacteria by permeabilizing and depolarizing the membrane. Furthermore, no resistance to the peptides was observed to develop in a hypermutant E. coli strain over 20 days. Collectively, our data indicate that adding Lys to the EMP-EM derivatives leads to increased positive charge, in turn likely yielding more effective electrostatic interactions with anionic bacterial membranes.^70–73 The structure-function-guided design approach based on Lys substitutions yielded peptide K¹³-2, which displayed increased antimicrobial activity, low toxicity against keratinocytes and HEK cells, and potent anti-infective properties in vivo compared with its predecessor WT2. To the best of our knowledge, this is the first time that the natural peptides EMP-EM1 and EMP-EM2 have been used as scaffolds for the rational design of optimized synthetic AMPs. Briefly, we demonstrated the ability of Ala scan screening and strategic single Lys substitutions to guide peptide design leading to AMPs with increased antimicrobial activity compared with their natural templates derived from venoms. The synthetic peptides designed here constitute active antibiotic scaffolds that warrant further development.

EXPERIMENTAL PROCEDURES

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Cesar de la Fuente-Nunez (cfuente@upenn.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

All data reported in this paper will be shared by the lead contact upon reasonable request. This work did not generate any code.

Solid-phase peptide synthesis, purification, and analysis

Peptides were purchased from AAPPTec, and synthesized by solid-phase peptide synthesis, using a fluorenylmethyloxycarbonyl (Fmoc) strategy on rink amide resin.

CD spectroscopy

CD experiments were performed on an Aviv CD spectrometer from the Biological Chemistry Resource Center (BCRC) of the University of Pennsylvania. CD spectra were recorded in three replicates at 25°C using a 0.25-mL quartz cuvette with 1.0-mm optical path length between 260 and 190 nm at 50 nm min⁻¹ and bandwidth of 0.5 nm. The concentration of all peptides was 50 μmol L⁻¹ and the measurements were performed in water and in a mixture of water and TFE 3:2 after recording the respective baselines.

Bacterial strains

The strains used in this work were E. coli ATCC 11775, E. coli JW2703 (hypermutant strain) ΔmutS::kan. (from the Keio collection) (kindly donated by Mark Goulian), P. aeruginosa PAO1, P. aeruginosa PA14, and S. aureus ATCC 12600.

MIC assays

The MIC assays were performed following the microdilution method.⁷⁴ Peptide solutions of 256 μmol L⁻¹ in Milli-Q sterile-filtered water were added to 96-well round-bottom plates, and a 2-fold serial dilution was performed to obtain peptide concentrations ranging from 128 to 2 μmol L⁻¹. Bacterial solutions at 5 × 10⁵ CFU mL⁻¹ in Luria-Bertani (LB) broth medium of E. coli ATCC 11775, S. aureus ATCC 12600, and P. aeruginosa (PAO1 and PA14) were added to the plates, and plates were incubated for 24 h at 37°C. After treatment, the optical density (OD) at 600 nm of the plates was measured on a Thermo Scientific Varioskan LUX fluorescence spectrophotometer to check bacterial growth inhibition and to compare results with those of untreated controls. Heatmaps obtained directly from OD measurements of the plates after treatment with all tested peptides and bacteria are shown in Figures S2 and S4. All MIC assays were performed in three replicates.

Bacterial killing experiments

Bacterial killing experiments to determine the MIC were performed according to Wiegand et al.⁷⁴ After 24 h of treatment, solutions corresponding to MIC and MIC/2 (identified by OD measurements) were collected and transferred to 96-well round-bottom plates, and serially diluted in 10-fold increments. Solutions were plated on LB agar plates (for E. coli ATCC 11775 and S. aureus ATCC 12600) and Pseudomonas Isolation Agar plates (for P. aeruginosa PAO1 and PA14), and cultures were incubated for 21 h at 37°C. Next, bacterial colonies were counted. The MBC was assessed by counting CFUs to confirm the MIC values of all peptides of this study against all bacterial strains tested. All assays were done in four replicates, including the controls.

NPN assay

The NPN assay for outer membrane permeabilization studies was performed based on the Hancock & Wong method.⁷⁵ P. aeruginosa PAO1 cells were grown to an OD at 600 nm of 0.5, centrifuged, and diluted ina5 mmol L⁻¹ sterile-filtered buffer solution N-2-hydroxyethylpiperazine-N′−2-ethanesulfonic acid (HEPES) with pH of 7.4 for 0.4 OD mL⁻¹. Bacteria cells were washed three times with HEPES solution and centrifuged for 5 min at 10,000 rpm. Solutions of WT1, K¹²-1, K¹³-1, WT2, K¹³-2, and PMB were added to a white 96-well plate at their corresponding MICs. Four microliters of a 5-mmol L⁻¹ stock solution of NPN in acetone was added to all wells containing peptide solution in the dark, to avoid photophysical decomposition. The bacterial solution, previously prepared in HEPES solution, was quickly added to the plate in the dark. The plate was read immediately and then every minute for 30 min, on a Thermo Scientific Varioskan LUX fluorescence spectrophotometer with the excitation wavelength set to 350 nm and emission wavelength set to 420 nm. All assays were done in three replicates, including the controls, which consisted of only HEPES solution, HEPES solution and NPN, HEPES solution and P. aeruginosa PAO1, and HEPES solution with both P. aeruginosa PAO1 and NPN.

DiSC₃-5 assay

The DiSC₃-5 assay for cytoplasmic membrane depolarization studies was performed according to Zhang et al.⁷⁶ P. aeruginosa PAO1 cells were grown to an OD at 600 nm of 0.5, centrifuged, and washed with a 5-mmol L⁻¹ sterile-filtered HEPES solution containing 20 mmol L⁻¹ of glucose with pH of 7.2. After washing, the supernatant was removed, and a bacterial solution was prepared in 5 mmol L⁻¹ of HEPES solution containing 20 mmol L⁻¹ of glucose and 0.1 mol L⁻¹ of KCl with pH 7.2. This bacterial solution was added to a black 96-well plate and read on the fluorescence spectrophotometer with the excitation wavelength set to 622 nm and the emission wavelength set to 670 nm. After that, 2 μL of a stock solution of DiSC₃-5 (0.1 mmol L⁻¹ in dimethyl sulfoxide) was added in the dark, and the plate was read every minute for 20 min. Solutions of the WT1, K¹²-1, K¹³-1, WT2, K¹³-2, and PMB peptides, at their corresponding MICs, were quickly added to the plate in the dark, and the plate was read immediately every minute for 60 min (until a plateau in the emission intensity was reached). After that, 5 μL of triton solution was added, and the plate was read every minute for 20 min (Figure S5). All assays were done in three replicates, including the controls, which consisted of only HEPES solution, and HEPES solution containing P. aeruginosa PAO1 and DiSC₃-5.

Synergy assays

After determining the MIC of each antibiotic and the four peptides tested (i.e., ciprofloxacin, ofloxacin, gentamicin, PMB, erythromycin, WT1, WT2, K¹²-1, and K¹³-1) for P. aeruginosa PAO1 (Figure S6), antibiotics and peptides were diluted using the microdilution technique to concentrations ranging from 2- to 0.03-fold MIC in different 96-well plates and then combined pairwise. Plates were incubated with 5 3 10⁵ CFU mL⁻¹ of P. aeruginosa PAO1 at 37°C for 24 h. All assays were done as three independent replicates. The OD at 600 nm was then measured (Figure S7), and the activity was studied using the FICI to evaluate synergy, considering FICI < 0.5 as synergistic, 0.5 < FICI < 1.0 as additive, 1.0 < FICI < 4.0 as indifferent, and FICI > 4.0 as antagonistic.^77–79

Resistance development assays

The MIC for E. coli JW2703 (hypermutant strain) ΔmutS::kan. (from the Keio collection) was determined with peptides WT1, K¹²-1, and K¹³-1 and the antibiotics ciprofloxacin and polymyxin B using a bacterial solution at 5 × 10⁵ CFU mL⁻¹ in nutrient broth (NB) medium, according to the procedure described above. After treatment, the OD at 600 nm was measured and the bacterial solutions containing the minimal concentrations of peptide or antibiotic in which the bacteria grew (at least 50% of the bacterial growth of the control) were collected, diluted in NB medium (1:100), and incubated overnight at 37°C with stirring (250 rpm). The remaining volume of all chosen solutions was stored in a round bottom 96-well plate containing sterile glycerol solution (50%) at −80°C. A 96-well plate was prepared containing peptide/antibiotic solutions with concentrations ranging from 8- to 0.25-fold the MIC. All pre-inoculums were diluted in NB medium (1:100) after incubation, added to the plate, and incubated overnight at 37°C. After treatment, OD at 600 nm was measured, and bacterial solutions containing the minimal concentrations of peptide or antibiotic in which the bacteria were able to grow were collected, diluted in NB medium (1:100), and incubated overnight at 37°C with stirring. The remaining volume of all chosen solutions was stored in a round-bottom 96-well plate containing sterile glycerol solution at −80°C. This procedure was repeated until bacterial resistance was observed (20 days). All tests were performed as three independent replicates.

Cytotoxicity assays

Human embryonic kidney 293T (HEK293T) cells were maintained in Dulbecco’s modified eagle medium (DMEM) supplemented with heat-inactivated 10% fetal bovine serum, and keratinocytes were maintained in Medium 154 supplemented with human keratinocyte growth supplement (HKGS, Gibco), Keratinocyte serum free medium (SFM), Combo Combination (Gibco), and with antibiotic/antimycotic solution (Gibco). Cells were seeded in 96-well tissue-culture-treated plates at a density of 5 × 10⁴ cells per well in their respective media. HEK293T cells were then incubated for 24 h and keratinocytes for 48 h, in 5% CO₂, at 37°C for cell adhesion. After adhesion, media were replaced and supplemented with concentrations of tested peptides ranging from 64 to 1 μmol L⁻¹ (or media without peptide as a control) in a final volume of 100 μL per well. After 24 h of incubation, 25 μL of activated 2,3-Bis-(2-Methoxy-4-Nitro-5-Sulfophenyl)-2H-Tetrazolium-5-Carboxanilide (XTT, Biotium) was added to each well and the plate was incubated for 5 h, in 5% CO₂, at 37°C. The XTT mitochondrial reduced product absorbance was measured at 460 nm and the background at 690 nm was subtracted. Cytotoxicity was determined as a percentage of the maximum value of cells without peptide compared with the minimum value of corresponding media without cells (% cell viability = 100 × [(X-MIN)/(MAX-MIN)]). Keratinocytes were purchased from the Penn Skin Biology and Diseases Resource-based Center (SBDRC).

Skin scarification mouse model

The methodology used is described in detail in Pane et al.⁵⁸ Briefly, the anti-infective activity of the peptides K¹²-1, K¹³-1, and K¹³-2 against P. aeruginosa PAO1 in a mouse model was assessed. CD-1 female mice (6 weeks old) were used and maintained in the University Laboratory Animal Resources (ULAR) at the University of Pennsylvania (protocol 806763). Four mice per group in each condition were used in two independent replicates to ensure accuracy. Mice were anesthetized with isoflurane, weighed, and their backs were shaved. A needle was then used to damage the stratum corneum and upper layer of the epidermis of the skin, causing a superficial linear skin abrasion. Fifty microliters of a bacterial solution at 10⁷ CFU mL⁻¹ in phosphate-buffered saline (PBS) of P. aeruginosa PAO1 was inoculated over the scratch in the back of the mice. After 1 h, peptide solutions in PBS of 32 μmol L⁻¹ for K¹³-2 and 16 μmol L⁻¹ for K¹²-1 and K¹³-1 were added to the infected area. After 2 days, mice from each group were killed and weighed, and the area of scarified skin was collected, homogenized using a bead beater for 20 min (25 Hz), and serially diluted for CFU quantification. This procedure was repeated after 4 days with mice from each group. The body weight of mice was monitored before and after 2 and 4 days of treatment to assess peptide cytotoxicity in vivo. The skin scarification mouse model (protocol number 806763) was revised and approved by the ULAR from the University of Pennsylvania.

Highlights.

Structure-function-guided design yielded peptide antibiotics with potent activity

Ala-Scan and lysine substitutions revealed antimicrobial hotspots

Net charge was shown as the most relevant feature for antimicrobial activity

INCLUSION AND DIVERSITY

One or more of the authors of this paper self-identifies as an underrepresented ethnic minority in their field of research or within their geographical location. One or more of the authors of this paper self-identifies as a gender minority in their field of research. We support inclusive, diverse, and equitable conduct of research.