Plantaricin A, Derived from Lactiplantibacillus plantarum, Reduces the Intrinsic Resistance of Gram-Negative Bacteria to Hydrophobic Antibiotics

ABSTRACT

The outer membrane of Gram-negative bacteria is one of the major factors contributing to the development of antibiotic resistance, resulting in a lack of effectiveness of several hydrophobic antibiotics. Plantaricin A (PlnA) intensifies the potency of antibiotics by increasing the permeability of the bacterial outer membrane. Moreover, it has been proven to bind to the lipopolysaccharide of Escherichia coli via electrostatic and hydrophobic interactions and to interfere with the integrity of the bacterial outer membrane. Based on this mechanism, we designed a series of PlnA1 analogs by changing the structure, hydrophobicity, and charge to enhance their membrane-permeabilizing ability. Subsequent analyses revealed that among the PlnA1 analogs, OP4 demonstrated the highest penetrating ability, weaker cytotoxicity, and a higher therapeutic index. In addition, it decelerated the development of antibiotic resistance when the E. coli cells were continuously exposed to sublethal concentrations of erythromycin and ciprofloxacin for 30 generations. Further in vivo studies in mice with sepsis showed that OP4 heightens the potency of erythromycin against E. coli and relieves inflammation. In summary, our results showed that the PlnA1 analogs investigated in the present study, especially OP4, reduce the intrinsic antibiotic resistance of Gram-negative pathogens and expand the antibiotic sensitivity spectrum of hydrophobic antibiotics in Gram-negative bacteria.

IMPORTANCE Antibiotic resistance is a global health concern due to indiscriminate use of antibiotics, resistance transfer, and intrinsic resistance of certain Gram-negative bacteria. The asymmetric bacterial outer membrane prevents the entry of hydrophobic antibiotics and renders them ineffective. Consequently, these antibiotics could be employed to treat infections caused by Gram-negative bacteria, after increasing their outer membrane permeability. As PlnA reportedly penetrates outer membranes, we designed a series of PlnA1 analogs and proved that OP4, one of these antimicrobial peptides, effectively augmented the permeability of the bacterial outer membrane. Furthermore, OP4 effectively improved the potency of erythromycin and alleviated inflammatory responses caused by Escherichia coli infection. Likewise, OP4 curtailed antibiotic resistance development in E. coli, thereby prolonging exposure to sublethal antibiotic concentrations. Thus, the combined use of hydrophobic antibiotics and OP4 could be used to treat infections caused by Gram-negative bacteria by decreasing their intrinsic antibiotic resistance.

INTRODUCTION

Gram-negative bacteria cause many types of infections and foodborne diseases that spread to humans via various mechanisms (1, 2). To make matters worse, certain Gram-negative bacteria show intrinsic resistance to multiple drugs because of the barrier imposed by their outer membranes and resistance acquired due to the inappropriate use of antibiotics (3–5). Therefore, improving the permeability of the outer membrane to the antibiotics might serve as an efficient method to tackle this issue (6, 7).

With the increasing instances of antibiotic resistance in Gram-negative bacteria, there is an urgent need for the development of novel effective drugs against them (3, 7, 8). However, discovering or designing effective antibiotics against Gram-negative bacteria is challenging (9, 10). It is known that cationic antimicrobial peptides bind to lipopolysaccharide (LPS) and phospholipids on the outer membranes of bacterial cells via electrostatic and hydrophobic interactions and damage the physical integrity of the inner membrane (11, 12). Furthermore, synergistic action of such peptides and antibiotics has been documented to decrease the dose of antibiotics and mitigate the development of antibiotic resistance (13, 14).

Plantaricin A (PlnA) is a cationic bacteriocin, produced by Lactiplantibacillus plantarum (15), which acts as an extracellular signal that triggers the bacteriocin production (16), in addition to possessing a bacterial strain-specific antagonistic activity (17). In our previous study (18), we classified plantaricins (PlnAs) into four categories in accordance with their amino acid sequences. PlnA3 has been reported to permeabilize eukaryotic cell membranes by interacting with anionic phospholipids and glycosylated membrane proteins (15, 19). Furthermore, compared with other plantaricins, such as PlnEF and PlnJK, PlnA3 has weaker antibacterial activity. PlnA3 (50 μM) alone has been shown to demonstrate antimicrobial activity that suppresses bacterial growth; however, the concentration was not bactericidal (20, 21). Thus, it was specifically the outer membrane-penetrating ability of PlnA and not its bactericidal activity that intrigued us and formed the basis of the present study.

Accordingly, in this study, we investigated the synergistic effect of PlnA1 with hydrophobic antibiotics against Gram-negative bacteria to assess the possibility of expanding the therapeutic efficacy spectrum of antibiotics. While the synergistic antimicrobial mechanism of PlnA1 and antibiotics remains unclear, we investigated the mechanism of antibiotic internalization in the presence of PlnA1. This was achieved by designing a series of PlnA1 analogs and assessing their synergistic antimicrobial effect with antibiotics on Escherichia coli to obtain a more efficient outer membrane-penetrating peptide.

RESULTS

PlnA1 exhibits the highest therapeutic index and synergy with hydrophobic antibiotics.

To determine the synergistic effects of PlnAs with hydrophobic antibiotics, the antibacterial activities and therapeutic indices (minimal hemolytic concentration [MHC]/MIC) of the four categories of PlnAs were examined. The results revealed that among all the categories of PlnAs, PlnA3 had the most potent antibacterial activity, with its MIC for E. coli being 6.25 μg/mL. However, PlnA1 demonstrated the highest therapeutic index (MHC/MIC) and the highest synergistic effect with erythromycin (Ery) among all the investigated PlnAs (Table 1). Consequently, PlnA1 was selected as the research object for the subsequent experiments, as the present study aimed to reduce the dosage of antibiotics and expand the antibacterial spectrum of existing antibiotics through combined treatment with PlnAs.

TABLE 1.

Synergistic effect on PlnAs and erythromycin

PlnA category	MIC (μg/mL) of:			MHC (μg/mL)a	MHC/MIC	FICIb
	PlnA	Ery	Ery + PlnA
PlnA1	25	500	32/6.25	480	12.8	0.314
PlnA2	25	500	64/6.25	480	12.8	0.378
PlnA3	12.5	500	64/6.25	60	3.2	0.628
PlnA4	50	500	64/25	60	1.2	0.628

^aWe defined the MHC as the highest peptide concentration that caused no detectable release of hemoglobin.

^bA FICI value of ≤0.5 indicates a synergistic effect, a value of >0.5 and ≤1 indicates an additive effect, a value of >1 and <2 indicates an indifferent effect, and a value of ≥2 indicates an antagonistic effect.

PlnA1 increases the outer membrane permeability by binding to LPS in E. coli.

Generally, the antibacterial mechanism of amphipathic antimicrobial peptides involves binding to and subsequent disruption of cell membranes (11). Thus, to study the effect of PlnA1 on E. coli, the changes in the cellular morphology of E. coli ATCC 35218 cells treated with various doses of PlnA1 were examined. The results demonstrated that the cell morphology of the E. coli ATCC 35218 cells was considerably altered when they were treated with one-fourth the MIC (6.25 μg/mL) of PlnA1 (Fig. 1). Furthermore, treating E. coli with 1× MIC of PlnA1 (25 μg/mL) resulted in the development of cavitation due to leakage of intracellular substances, while pronounced cavitation and membrane rupture were detected when the cells were treated at 4× to 8× MIC. Additionally, the outer membrane permeability of E. coli ATCC 35218 was increased significantly (P < 0.05) upon treatment with one-fourth the MIC (6.25 μg/mL) of PlnA1 (Fig. 2a).

FIG 1.

Cell morphology changes in Escherichia coli in the presence of PlnA1. E. coli ATCC 35218 cells were treated with 0 to 200 μg/mL of PlnA1. The morphology changes were observed using transmission electron microscopy.

FIG 2.

PlnA1 binds to the lipopolysaccharides of the outer membrane in Escherichia coli. E. coli ATCC 35218 was used as the indicator strain. (a) Permeability of the outer membranes of E. coli cells incubated with PlnA1 (0 to 200 μg/mL). (b) Outer membrane permeability of E. coli incubated with PlnA1 (12.5 and 25 μg/mL) after removal of the outer membrane proteins or polysaccharides via treatment with trypsin (1 mg/mL) or glycosidase (10 mU/mL), respectively. (c) Outer membrane permeability of E. coli incubated with PlnA1 (25 μg/mL) and magnesium ion (100 mM), calcium ion (100 mM), cholesterol (Cho) (20 μg/mL), lecithin (20 μg/mL), lipopolysaccharide (LPS; 10 to 160 μg/mL), or peptidoglycan (PGN; 20 μg/mL). (d) Liposomal leakage rate from the liposomes and LPS-liposomes treated with PlnA1 (0 to 500 μg/mL). (e) Endotoxin LPS levels in the medium after the treatment of E. coli ATCC 35218 cells with PlnA1 (12.5 and 25 μg/mL), EDTA (30 μg/mL), or polymyxin B (8 μg/mL). (f) Detection of binding between LPS (20 μg/mL) and PlnA1 (25 μg/mL) using high-performance liquid chromatography (HPLC). *, P < 0.05.

To identify the specific PlnA1-binding moieties on the outer membrane, the E. coli cells were treated with trypsin or glycosidase to remove the outer membrane proteins and surface polysaccharides, respectively. Subsequently, experiments demonstrated that trypsin treatment did not affect the outer membrane-penetrating ability of PlnA1, whereas treatment with glycosidase reduced the penetrating ability of PlnA1 in E. coli by 95% (Fig. 2b). These findings suggest that polysaccharides might serve as the binding targets of PlnA1 on the outer cell membrane in E. coli. Additionally, magnesium ions, calcium, lecithin, and LPS, as well as pH changes (see Fig. S1 in the supplemental material), evidently reduced the outer cell membrane-penetrating ability of PlnA1 in E. coli (Fig. 2c). These results indicate that PlnA1 competitively binds with divalent cations compared to LPS via electrostatic interaction, resulting in the disruption of the LPS layer on the outer membrane of E. coli cells. Likewise, the liposome leakage studies with or without LPS on the surface of liposomes revealed that at a 50% leakage rate, the effective concentration of PlnA1 in liposomes with LPS was reduced by 50-fold, compared with that in liposomes without LPS. These results thus suggest that liposomes with LPS were more sensitive to PlnA1 than liposomes without LPS (Fig. 2d). Furthermore, the release of LPS from E. coli ATCC 35218 increased 6-fold when the cells were incubated with PlnA1, which was higher than that observed in the cells incubated with EDTA and polymyxin B (Fig. 2e). Moreover, the native binding between PlnA1 and LPS was observed using high-performance liquid chromatography (HPLC) (Fig. 2f; also, see Fig. S2). Accordingly, the peak of the PlnA1-LPS conjugate was detected at 9.827 min, while the molecular weight was found to have increased to 73 kDa based on a standard molecular weight curve (Fig. S3). These data suggested that PlnA1 binds to LPS via electrostatic and hydrophobic interactions, which subsequently destroys the bridges between the LPS moieties and diminishes the integrity of outer cell membrane in bacteria.

Absence of LPS on the surface of E. coli reduces the outer membrane-penetrating ability of PlnA1.

To further confirm the effect of LPS on the mode of action of PlnA1, the LPS synthesis genes of E. coli, such as lpxA, waaC, and waaY, were knocked out (Fig. 3a; Fig. S4). Subsequent assessment of endotoxin levels revealed that the levels of free LPS in E. coli ΔlpxA and E. coli ΔwaaC strains were mitigated by 88 to 90%, which led to a 75 to 80% decrease in the outer membrane-penetrating ability of PlnA1 (Fig. 3b and c). In contrast, overexpression of lpxA and waaC increased the endotoxin levels by 160 to 180% and restored the penetrating ability of PlnA1 in E. coli ΔlpxA and E. coli ΔwaaC strains, respectively (Fig. 3b). In addition, mcr-1, which reduces the negative charge of LPS, was expressed in E. coli. Subsequent assessment of outer membrane-penetrating ability of mcr-1-overexpressing E. coli cells demonstrated that, consistent with the prediction, the penetrating ability of PlnA1 reduced by 50% and 62.5%, respectively (Fig. 3c). In summary, our findings confirmed that PlnA1 could bind to LPS on the outer membrane of E. coli cells via electrostatic interactions, thereby interfering with the structure of the outer membrane and increasing its permeability.

FIG 3.

Gene knockout and overexpression assays indicate the binding site between PlnA1 and LPS. (a) Schematic representation of LPS and the genes involved in its synthesis. UDP-GlcNac, UDP N-acetylglucosamine; Hep, heptulose; Kdo, keto-deoxyoctulosonate; PEtN, phosphoethanolamine. (b) Differences in endotoxin levels between lpxL, waaC, and waaY knockout and mcr-1-, lpxL-, waaC-, and waaY-overexpressing Escherichia coli strains. (c) Effect of LPS modification on the outer membrane permeability after incubation with PlnA1 (6.25 μg/mL). *, P < 0.05.

Designing novel PlnA1-derived peptides with higher outer membrane-penetrating ability.

Our circular dichroism studies have shown that the percentage of α-helix in PlnA1 was 53.18% in 50% trifluoroethanol (Fig. S5). This finding suggests that PlnA1 is an amphipathic molecule with basic amino acids distributed on one side and hydrophobic amino acids on the other (Fig. 4b and c). Moreover, studies involving alteration in the charge, hydrophobicity, structure, and amphipathicity of PlnA1 (Fig. 4a) showed that the permeability of the outer membrane was significantly determined by the charge, hydrophobicity, and α-helix of the peptide (P < 0.05) (Fig. 5b to d). Furthermore, the penetrating ability of the PlnA1 analogs was found to be decreased in peptides with reduced positive charges (such as K43D, K50D, and K55D) (Table 2). Accordingly, OP1, OP2, and OP4 were designed based on the fitting equation, with different positive charges, α-helices, and amphipathicities. Further analysis of OP4 revealed that it exhibits a more amphiphilic structure than that of PlnA1 and that its hydrophobic amino acids are distributed oppositely to the basic amino acids, as shown in the helical wheel diagram based on the fitting equation (Fig. 4d). Moreover, the E. coli outer membrane permeability was determined to be increased by 2-fold in the presence of OP4 (0.78 μg/mL) compared with that in the presence of PlnA1 (6.25 μg/mL). Finally, the docking model of LPS and OP4 confirmed that PlnA1 binds to LPS via electrostatic interactions and interferes with the structure of the LPS (Fig. S6).

FIG 4.

Characteristics of PlnA1 and OP4. (a) PlnA1 analogs were constructed by alanine mutation, introducing a positive charge, and altering hydrophobicity. (b) Spatial structure of PlnA1 and OP4 predicted by Phyre2. (c) Helical wheel diagram of PlnA1. Its polarity angle is 120°. (d) Helical wheel diagram of OP4. Its polarity angle is 160°. Hydrophobic and basic amino acids are in red and blue, respectively, while other hydrophilic amino acids are in green.

FIG 5.

Key factors determining the penetrating ability of PlnA1. Data were analyzed using the multiple linear regression tool of GraphPad Prism 8.0.2. (a) Relationship between the outer membrane permeability and PlnA1 characteristics. (b, c, and d) Three-dimensional surfaces showing the relationships between permeability, hydrophobicity, helix, and charge.

TABLE 2.

Penetrating ability of designed PlnA1 analog peptides

PlnA1 analog	Charge	Hydrophobicity	α-Helix (%)	OM permeability (%)
PlnA1	6	−0.37	53.33	100
N38D	5	−0.37	56.66	80
T44E	5	−0.463	60	60
K43D	4	−0.357	53.33	60
K50D	4	−0.357	53.33	40
K55D	4	−0.357	50	40
R42D	4	−0.337	53.33	80
N47L	6	−0.127	53.33	80
N31A	6	−0.193	53.33	100
S51I	6	−0.193	56.66	110
V40S	6	−0.537	56.66	110
I32N	6	−0.637	53.33	100
L52N	6	−0.613	46.66	50
L34N	6	−0.613	36.66	50
L34F	6	−0.403	53.33	90
F48A	6	−0.403	56.66	90
N31A+Y36A+G37M	6	−0.013	70	180
OP1	4	−0.483	46.66	60
OP2	3	−0.693	50	140
OP4	8	−0.496	66.66	200

Synergistic antibacterial effect of OP4 on E. coli in combination with different antibiotics.

To determine the effect of OP4 in combination with different antibiotics against E. coli, the checkerboard assay was performed using E. coli ATCC 35218. Subsequent results showed that OP4 enhanced the potency of Ery by 62.5-fold at 0.78 μg/mL, fusidic acid by 64-fold at 1.56 μg/mL, clindamycin by 128-fold at 1.56 μg/mL, quinolones (i.e., ciprofloxacin [CIP] and norfloxacin) by 5- to 7-fold, and trimethoprim by 16-fold (Table 3).

TABLE 3.

MICs of antibiotics with or without PlnA1 analogs against E. coli^a

Antibiotics	MIC (μg/mL)	MIC (μg/mL) (in combination) of:		FICIb	MIC (μg/mL) (in combination) of:		FICIb
		Antibiotic	OP4		Antibiotic	PlnA1
Ampicillin	4	4	3.13	2.0	4	25	2.0
Amoxicillin	2	2	3.13	2.0	2	25	2.0
Erythromycin	500	8	0.78	0.266	32	6.25	0.314
Streptomycin	8	8	1.56	1.5	8	25	2.0
Gentamycin	2	2	1.56	1.5	2	25	2.0
Kanamycin	4	4	1.56	1.5	4	12.5	1.5
Tetracycline	2	0.25	0.78	0.375	0.5	12.5	0.75
Oxytetracycline	2	0.25	0.78	0.375	0.5	6.25	0.5
Lincomycin	500	500	3.13	2.0	250	25	1.5
Chloramphenicol	4	4	1.56	1.5	4	25	2.0
Polymyxin B	8	4	1.56	1.0	4	12.5	1.5
Vancomycin	125	125	3.13	2.0	125	12.5	1.5
Ciprofloxacin	0.06	0.008	0.78	0.38	0.016	6.25	0.5
Norfloxacin	0.04	0.008	0.78	0.445	0.01	6.25	0.5
Trimethoprim	8	0.5	0.78	0.3125	1	6.25	0.375
Triclosan	2	1	0.78	0.75	1	12.5	1.0
Novobiocin	62.5	7.81	0.78	0.375	7.81	6.25	0.375
Fusidic acid	125	15.625	0.39	0.25	15.625	3.125	0.25
Clindamycin	250	1.95	0.78	0.2578	1.95	12.5	0.5078
Rifampicin	3.125	0.195	0.39	0.1874	0.195	6.25	0.3124

^aThe indicator strain used was E. coli ATCC 35218. The MIC of PlnA1 was 25 μg/mL, and the MIC of OP4 was 3.125 μg/mL.

^bFICI values of <0.5, >0.5 to 1, >1 to <2, and ≥2 indicate synergism, additive effects, indifference, and antagonism, respectively.

Furthermore, except E. coli, the OP4 treatment enhanced the potency of Ery by 20- to 80-fold against a variety of Gram-negative pathogens, such as enterohemorrhagic E. coli O157:H7 CICC 21530, Yersinia pseudotuberculosis CMCC 52225, Vibrio parahaemolyticus ATCC 33847, multidrug-resistant Salmonella CDC-1, Pseudomonas aeruginosa ATCC 27853, Pseudomonas fluorescens AS 3.6452, and Klebsiella pneumoniae CICC 21519 (Table 4). These results indicate that OP4 increases the outer membrane permeability of a variety of Gram-negative bacteria and enhances the potency of hydrophobic antibiotics.

TABLE 4.

OP4 enhances the potency of erythromycin against Gram-negative pathogens

Pathogena	MIC (μg/mL) of:		Fold difference
	Ery	Ery with OP4
Enterohemorrhagic E. coli O157:H7 CICC 21530	800 ± 25	10 ± 2	80
Yersinia pseudotuberculosis CMCC 52225	800 ± 25	10 ± 2	80
Vibrio parahaemolyticus ATCC 33847	400 ± 10	5 ± 1	80
Salmonella enterica serovar Typhimurium ATCC 14028	600 ± 25	10 ± 2	60
Salmonella enterica serovar Bredeney CDC-1	1,200 ± 50	15 ± 3	80
Pseudomonas aeruginosa ATCC 27853	1,000 ± 50	50 ± 5	20
Pseudomonas fluorescens AS 3.6452	600 ± 25	15 ± 5	40
Klebsiella pneumoniae CICC 21519	800 ± 50	25 ± 5	32

^aCICC, China Center of Industrial Culture Collection; CMCC, China Center of Medicine Culture Collection; ATCC, American Type Culture Collection; AS, China General Microbiological Culture Collection Center. Salmonella enterica serovar Bredeney CDC-1, which was available in our lab, is a multidrug-resistant pathogen described previously by Ju et al. (56).

OP4 enhances the accumulation of Ery and curtails the development of antibiotic resistance in E. coli.

To verify the synergy between OP4 and antibiotics, we measured the concentration of antibiotics that accumulated in the E. coli cells after treatment with different antibiotics. The results showed that the concentration of antibiotics in the wild-type E. coli cells was significantly increased (P < 0.05) after incubation with OP4. For instance, the intracellular Ery concentration was found to be increased by 10-fold, while that of tetracycline, CIP, and trimethoprim was determined to be increased by approximately 2-fold (Fig. 6a). Furthermore, decreasing the LPS level or negative charge on the cell surface reduced the intracellular accumulation of Ery by 43%, 90%, 76%, and 75% in E. coli mcr-1, E. coli ΔlpxA, E. coli ΔwaaC, and E. coli ΔwaaY cells, respectively. In contrast, increased levels of LPS, such as those evident in E. coli ΔlpxA/lpxA and E. coli ΔwaaC/waaC cells, promoted the intracellular accumulation of Ery up to 129% and 124%, respectively (Fig. 6e; Fig. S7).

FIG 6.

OP4 promotes the uptake of antibiotics by Escherichia coli. E. coli ATCC 35218 was used as the indicator strain. (a) Antibiotic accumulation by E. coli, as quantified via HPLC, with or without OP4 treatment (0.78 μg/mL). (b) Erythromycin accumulation in E. coli after LPS modification, with or without OP4 treatment (0.78 μg/mL). (c) Development of antibiotic resistance by E. coli incubated with sublethal concentrations of erythromycin or ciprofloxacin (CIP) with or without OP4 treatment (0.78 μg/mL). *, P < 0.05.

Additionally, the antibiotic resistance of E. coli did not develop when the cells were treated with Ery in combination with OP4. However, a 4-fold increase in antibiotic resistance was detected in E. coli after 30 days when the cells were treated with Ery alone. Similarly, the CIP resistance of E. coli in the presence of OP4 (OP4+CIP) was increased by 2-fold, whereas that of the control group was increased by 16-fold (Fig. 6). These results suggest that the synergy between OP4 and Ery or CIP not only decreased the effective antibiotic concentration but also curtailed the development of antibiotic resistance in E. coli.

OP4 exerts a low cytotoxicity and hemolysis and enhances the in vivo efficacy of Ery against E. coli in mice.

In the present study, the MICs of PlnA1 and OP4 against E. coli ATCC 35218 were determined to be 25 and 3.125 μg/mL, respectively, while they increased the outer membrane permeability at 6.25 and 0.78 μg/mL, respectively. Furthermore, these concentrations were much lower than those causing hemolysis and toxicity in HepG2 and Caco-2 cells (Fig. 7).

FIG 7.

Toxicity assessment of combined PlnA1 and OP4 treatment. (a) Hemolytic activity of combined PlnA1 and OP4 treatment at different concentrations on sheep red blood cells. (b) Average cytotoxicity of combined PlnA1 and OP4 treatment at different concentrations in HepG2 and Caco-2 cells.

To further investigate the synergy between OP4 and antibiotics, Ery and OP4 were used either alone or in combination to treat the murine sepsis caused by E. coli ATCC 35218 (Fig. 8). After mice were injected with E. coli in their abdominal cavities, the mortality rate reached 40% within 24 h and 60% on day 2. Similarly, the mortality of mice in the Ery treatment group reached 50% on day 2, indicating that 100 μg/g body weight (BW) of Ery was ineffective in inhibiting the growth of E. coli. In contrast, the mortality of mice in the Ery+OP4 treatment group was significantly reduced to 20% (P < 0.05) on day 2. These results showed that 10 μg/g Ery plus 10 μg/g OP4 could effectively treat E. coli-induced abdominal infections in mice (Fig. 9a). Moreover, Ery+OP4 treatment, but not Ery treatment, significantly alleviated the bacterial loads in the blood, heart, liver, spleen, lung, and kidneys of the infected mice (Fig. 9b to g). For instance, the bacterial loads in the mouse blood, heart, spleen, and kidneys treated with Ery+OP4 decreased by more than 10-fold, compared to those in the untreated infection group on day 1, and decreased by more than 5-fold on day 7. Subsequently, inflammation-related cytokines, including C-reactive protein (CRP), tumor necrosis factor alpha (TNF-ɑ), interleukin 1β (IL-1β), IL-6, and nuclear factor κB (NF-κB), were detected in plasma using the respective enzyme-linked immunosorbent assay (ELISA) kits (Fig. 9h to l). Although the level of inflammatory factors in mice increased significantly after E. coli infection, treatment with Ery did not alleviate the occurrence of inflammation. However, Ery+OP4 treatment significantly reduced (P < 0.05) the level of inflammatory factors, thereby suggesting that OP4 increased the potency of Ery in E. coli-infected mice by at least 10-fold.

FIG 8.

Schematic representation of the time frame of the murine sepsis study.

FIG 9.

OP4 improves the inhibitory activity of erythromycin (Ery) against Escherichia coli and reduces infection-associated inflammation in mice. A total of 10⁷ CFU of E. coli ATCC 35218 were intraperitoneally injected into BALB/c mice, and 100 μg/g Ery or 10 μg/g Ery plus 10 μg/g OP4 (Ery+OP4) was intraperitoneally injected into mice 2 h after infection. (a) The survival curve of mice was plotted using GraphPad. The bacterial loads in the (b) blood, (c) heart, (d) liver, (e) spleen, (f) lungs, and (g) kidneys were enumerated on LB agar plates at days 1 and 7. The levels of inflammation-related cytokines, including (h) C-reactive protein (CRP), (i) tumor necrosis factor alpha (TNF-α), (j) interleukin 1β (IL-1β), (k) IL-6, and (l) nuclear factor κB (NF-κB) were determined using ELISA. Different lowercase letters indicate significant differences (P < 0.05), and uppercase letters indicate significant differences (P < 0.05) at day 7 (b to g). Negative and control groups consisted of mice without any infection and infected mice without any treatment, respectively.

DISCUSSION

Antibiotic resistance is one of the biggest public health challenges of our time. In the United States alone, annually, at least 2.8 million people get an antibiotic-resistant infection and at least 35,000 people die from such infections (8). It has been estimated that the number of deaths due to drug-resistant infections will exceed 10 million per year by 2050 (22). The inappropriate use of antibiotics has been considered the major reason for the emergence of antibiotic resistance, which is known to exacerbate clinical bacterial infections. Moreover, some clinical isolates exhibit cross-resistance to various drugs (23, 24). Accordingly, reducing the use of antibiotics and developing novel antibacterial agents that are not prone to resistance is considered an optimal strategy to solve this problem (25, 26). Additionally, the PlnA1 analogs OP4 significantly increased the permeability of the E. coli outer membrane at a concentration that is one-fourth its MIC (0.78 μg/mL) and enhanced the potency of hydrophobic antibiotics (Ery, tetracycline, CIP, norfloxacin, and trimethoprim) against various pathogenic bacteria, such as enterohemorrhagic E. coli O157:H7, Y. pseudotuberculosis (which causes mesenteric lymphadenitis), V. parahaemolyticus (which causes food poisoning), and multidrug-resistant Salmonella.

Furthermore, the dense, hydrophobic, asymmetric outer membrane of Gram-negative bacteria serves as an important barrier against the entry of antibiotics into the cells, which confers intrinsic antibiotic resistance to bacteria. Discovering or designing antibiotics against Gram-negative bacteria is a challenge. Since the discovery of quinolone, a specific drug for Gram-negative bacteria, in 1968, no new drugs have been noted to be effective against Gram-negative bacteria in the past 50 years. In 2017, GlaxoSmithKline screened 500,000 prospective compounds but did not get the expected results. To solve this problem, researchers are screening novel compounds modified from existing antibiotics or combining multiple antibacterial drugs (cocktail therapy) to combat Gram-negative bacterial infections. Thus, combination therapy has emerged as a useful approach to overcome these difficult-to-treat infections. Current combination therapeutic strategies have evolved to combine clinical antibiotics with other antimicrobial agents (e.g., β-lactams plus aminoglycosides, amoxicillin plus clavulanate, and piperacillin plus sulbactam), antimicrobial peptides (e.g., M33-D, TAT, LL-37, hBD-3, and nisin) (27–29), or natural compounds (such as epigallocatechin gallate plus β-lactams and berberine plus azithromycin) (30–32). Moreover, PlnA has been found to exhibit strain-specific antagonistic activity against lactic acid bacteria (17). In the present study, PlnA1 increased the permeability of the E. coli outer membrane and demonstrated a weak hemolytic activity, low cytotoxicity, and high plasma stability (Fig. S8), suggesting that PlnA1 can be used as a potential permeabilization agent for the outer membrane of Gram-negative bacteria. The combination of antimicrobial drugs has many potential benefits, such as a broader therapeutic spectrum, reduced toxicity, lower likelihood of emergence of acquired resistance, and synergistic or additive interactions (33–35).

Similar to the results of Sand et al., wherein PlnA bound to glycosylated membrane proteins on the surface of the cell membrane (15), we confirmed that PlnA1 binds to the LPS of E. coli through electrostatic and hydrophobic interactions. However, the interaction of PlnA1 with the outer membrane of bacterial cells is different from that with the binding sites of polymyxin B and nisin, which interfere with the outer membrane by binding to lipid A (36) and lipid II, respectively, to increase the permeability of the outer membrane (37, 38). In addition, OP4 enhanced the antibacterial activity of Ery against E. coli and alleviated the infection-associated inflammation in mice in a murine sepsis model. Therefore, OP4 was identified as a potential drug delivery vector that could overcome the challenge of increasing resistance of bacteria to conventional antibiotics.

Many cell-penetrating peptides have been developed to transport therapeutic substances from the cell membrane to their target site. For instance, pyrrhocoricin, comprising 20 amino acids, has been used as a drug delivery system to deliver peptide antigens (NPK^d peptide) into dendritic cells and human fibroblasts. Additionally, lactoferricin (39), Bac7 (40), SynB1/PG-1 (41), LL-37 (42), and buforin II (BF2d) (43) have been employed to deliver drugs into various of cells (44). However, outer membrane-penetrating peptides have rarely been reported previously (45). For instance, PL-5 is an α-helical antimicrobial peptide (AMP) that reportedly synergizes with conventional antibiotics to improve the in vitro and in vivo antibacterial activity against both Gram-positive and Gram-negative bacteria (45). In the present study, the PlnA1 analogs, especially OP4, increased the permeability of the outer membrane and demonstrated synergy with hydrophobic antibiotics against Gram-negative bacteria. Furthermore, the combined use of OP4 and antibiotics hindered the development of drug resistance in E. coli cells. These findings indicate that such combination treatments may be effective in controlling the development of drug resistance in bacteria, thereby helping to combat bacterial infections.

MATERIALS AND METHODS

Strains and growth conditions.

Indicator bacteria were cultured in Mueller-Hinton (MH) broth at 35°C. The E. coli strains (Table 5) were cultured in Luria-Bertani (LB) medium at 37°C or at 30°C (when harboring the vector pCas).

TABLE 5.

Bacterial strains and vectors used in the study

Strain or vector	Description	Source or reference
E. coli strains and genotypes
DH5α	F− ϕ80dlacZΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 hsdR17(rK− mK+) phoA supE44 λ− thi-1 gyrA96 relA1; host for subcloning of genes and vectors	Vazyme Co., Ltd.
ATCC 35218	Beta lactamase producer; does not contain the stx1 or stx2 genes for Shiga toxin production	ATCC
ΔwaaC	ATCC 35218 with deletion of waaC	This work
ΔwaaY	ATCC 35218 with deletion of waaY	This work
ΔlpxA	ATCC 35218 with deletion of lpxA	This work
ΔwaaC/waaC	E. coli ΔwaaC with pUC57-waaC	This work
ΔwaaY/waaY	E. coli ΔwaaY with pUC57-waaY	This work
ΔlpxA/lpxA	E. coli ΔlpxA with pUC57-lpxA	This work
Vectors
pUC57	Expression vector of E. coli	Thermo Scientific
pUC57-mcr1	pUC57 with phosphoethanolamine-lipid A transferase mcr-1 (gene ID 39727008)	This work
pUC57-waaC	pUC57 with waaC	This work
pUC57-waaY	pUC57 with waaY	This work
pUC57-lpxA	pUC57 with lpxA	This work
pTargetF	sgRNA vector	51
pCas	Gene knockout vector with Cas9	51
pTargetF-waaC	Expresses sgRNA-waaC	This work
pTargetF-waaY	Expresses sgRNA-waaY	This work
pTargetF-lpxA	Expresses sgRNA-lpxA	This work

Pfu DNA polymerase, T4 DNA ligase, and T4 PNK were purchased from TaKaRa Biotechnology (Dalian) Co., Ltd., Liaoning, People’s Republic of China. The primers were synthesized by GenScript (Nanjing) Co., Ltd. (Jiangsu, People’s Republic of China). DOPC (1,2-dioleoyl-sn-glycero-3-phosphocholine) and DOPE (1,2-dioleoyl-sn-glycero-3-phosphatidyl-ethanolamine) were purchased from Sigma-Aldrich (MO, USA).

Properties of PlnA.

PlnAs were synthesized using chemical solid phase by GenScript, and their purity was determined to be >85%. The amino acid sequences of PlnAs are provided in Table 6.

TABLE 6.

Amino acid sequences of PlnAs^a

PlnA	Sequence	Theoretical MW
PlnA1	QFKNISLMYGNNVSRKTLTNFFKSLIKKI	3,438.11
PlnA2	KFQSVSLMYGNNVSRKTLTKFFKSLTKR	3,351.93
PlnA3	KSSAYSLQMGATAIKQVKKLFKKWGW	3,027.51
PlnA4	KTKTISLMSGLQVPHAFTKLLKALGGHH	3,056.64

^aAll sequences have N-terminal acetylation.

The amino acid composition, isoelectric point, charge, and hydrophobicity of PlnAs were analyzed using ProtParam (https://web.expasy.org/protparam). The amphipathicity and helical wheel diagrams of the peptides were drawn using BioEdit (https://bioedit.updatestar.com/). The spatial structures were predicted using Phyre2 (http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index). The interaction between LPS and PlnA was docked by AutoDock Tools 1.5.6. The PlnA analogs were then designed by changing the charge, hydrophobicity, and amphipathicity of the parent molecule, as described in Table 1.

Detection of MIC.

Indicator strains were cultured in MH broth to mid-logarithmic phase. The MICs of PlnA1 analogs and the PlnA1/OP4-antibiotic mixture were tested using the CLSI broth microdilution method (46) as follows. The concentration of indicator bacteria was adjusted to 5 × 10⁵ CFU/mL in 96-well plates, and the organisms were incubated at 35°C for 18 h after addition of gradient-diluted antimicrobial agents. The breakpoints were determined as described by EUCAST (https://www.eucast.org/clinical_breakpoints/). The MIC was considered the lowest concentration of an antibiotic (or peptide) at which bacterial growth was completely inhibited; i.e., the change in OD₆₀₀ was less than 5% compared with that of the positive control (50 μg/mL kanamycin).

Measurement of hemolytic activity.

A series concentration (0 to 500 μg/mL) of peptides was added to 1% human erythrocytes in phosphate-buffered saline (PBS; 80 mM NaCl, 43 mM Na₂HPO₄, 11 mM KH₂PO₄ [pH 7.0]) and maintained at 37°C for 1 h. The optical density at 562 nm (OD₅₆₂) of each supernatant was determined with a spectrophotometer. Triton X-100 solution (0.1%) and PBS were used as the positive and negative controls, respectively. For the purpose of this study, the MHC was defined as the highest peptide concentration that caused no detectable release of hemoglobin, i.e., the value of the OD₅₆₂ for the peptide solution was less than 5% compared with that of the negative control.

Preparation of electron microscopy samples.

The E. coli cells were cultured as described above and treated with different concentrations of PlnA1 (0, 6.25, 12.5, 25, 100, and 200 μg/mL) for 1 h. The cells were then harvested, washed twice with PBS (50 mM, pH 7.2), and fixed with 2.5% glutaraldehyde (Sigma-Aldrich). The samples were then sectioned and observed using a transmission electron microscope (47, 48).

Permeability assessment of outer membranes.

The permeability of the bacterial outer membrane was determined using N-phenyl-1-naphthylamine (NPN; Sigma-Aldrich) as follows (49). E. coli ATCC 35218 cells were cultured at mid-logarithmic phase, and the OD₆₀₀ of the suspension culture was adjusted to 0.05 using HEPES buffer (5 mM, pH 7.2; Sigma-Aldrich). A series of concentrations (0 to 100 μg/mL) of PlnA1 were added to the suspension cultures, and cultures were incubated at 37°C for 1 h. Subsequently, NPN was added to the mixtures to a final concentration of 10 μM. The excitation and emission wavelengths for the estimation of NPN were set to 350 and 420 nm, respectively.

In vitro PlnA1 binding experiments.

The E. coli ATCC 35218 cells were cultured at mid-logarithmic phase and adjusted to 5 × 10⁵ CFU/mL using HEPES buffer (5 mM, pH 7.2).

(i) Substrates.

PlnA1 (final concentration 25 μg/mL) was mixed with magnesium ion (100 mM), calcium ion (100 mM), cholesterol (Cho; 20 μg/mL; Sigma-Aldrich), lecithin (20 μg/mL; Sigma-Aldrich), LPS (10 to 160 μg/mL; Sigma-Aldrich), or peptidoglycan (20 μg/mL; Sigma-Aldrich) and incubated with E. coli ATCC 35218 cells at 37°C for 1 h for assessment of the outer membrane permeability using the fluorescence of NPN.

(ii) Outer membrane surface treatment.

E. coli ATCC 35218 cells were treated with trypsin (10 mg/mL; Thermo Fisher Scientific, Inc., MA, USA) or N-glycosidase F (10 mU/mL; New England Biolabs, MA, USA) at 37°C for 1 h. The fluorescence of NPN was then monitored as mentioned above. The cells were then centrifuged at 5,000 × g for 5 min and suspended in PBS; the changes in the permeability of the E. coli outer membrane were detected after incubation with PlnA1 (12.5 and 25 μg/mL).

(iii) LPS release from outer membranes.

E. coli ATCC 35218 cells were incubated with HEPES buffer (5 mM, pH 7.2), PlnA1 (12.5 and 25 μg/mL), EDTA (30 μg/mL), or polymyxin B (8 μg/mL) at 37°C for 1 h. The level of endotoxin (LPS) in the supernatant was quantified using a ToxinSensor chromogenic LAL endotoxin assay kit (GenScript).

(iv) HPLC.

The interaction of LPS (20 μg/mL) and PlnA1 (25 μg/mL) was monitored using an Agilent Bio Sec-5 column (5 μm, 7.8 mm by 300 mm; Agilent Technologies, Inc., CA, USA), and an acetonitrile-water mixture (95:5, containing 0.1% [vol/vol] trifluoroacetic acid [TFA]) was used as the mobile phase at a flow rate of 1.5 mL/min. The interaction detection wavelength was set at 220 nm.

Preparation of liposomes for liposome leakage assays.

Small unilamellar vesicles (SUVs) were prepared using sonication, as described previously (50). Briefly, the DOPC-DOPE mixture (0.5 mg–0.5 mg; Sigma-Aldrich) was dissolved in chloroform and allowed to form a film on the bottom of a round-bottom flask after evaporation. Subsequently, it was suspended in 1.0 mL of HEPES buffer (50 mM, pH 7.2, and 70 mM calcein). Then, it was sonicated (at 30% power, with an intermittent on [1 s]/off [2 s] cycle) until the liquid was cleared to form SUVs. Finally, these SUVs were purified using Sephadex G-50 to remove free calcein.

Additionally, the LPS-liposome SUVs (LPS/lipid ratio, 0.01 mg:1 mg; Sigma-Aldrich) were prepared using the sonication method described above. Subsequently, the liposomes or LPS-liposome SUVs were treated with different concentrations of PlnA1 (0 to 500 μg/mL), and the leakage of liposomes was monitored at excitation and emission wavelengths of 490 nm and 515 nm, respectively. Triton X-100 (1%) solution was used as a positive control.

Modification of LPS on the outer membrane of E. coli cells. (i) Membrane charge modification.

The mcr-1 (phosphoethanolamine-lipid A transferase; gene ID 39727008) gene was synthesized by GenScript, cloned into the pUC57 vector with a lac promoter and lac operator, and then transformed into E. coli ATCC 35218 cells. The outer membrane permeability of E. coli transformants was monitored after induction with isopropyl-β-d-thiogalactopyranoside (0.1 mM).

(ii) LPS-modified strains.

The LPS synthesis-related genes lpxA (UDP-N-acetylglucosamine acyltransferase; gene ID 944849), waaC (ADP-heptose-LPS heptosyltransferase; gene ID 948136), and waaY (LPS core heptose[II] kinase; gene ID 948145) were deleted according to the method described by Jiang et al. (51) (Table 7). Subsequently, the lpxA, waaC, and waaY genes were cloned into pUC57 and transformed into E. coli ΔlpxA, E. coli ΔwaaC, and E. coli ΔwaaY, respectively. Finally, the outer membrane permeabilities of E. coli ΔlpxA, E. coli ΔwaaC, E. coli ΔwaaY, E. coli ΔlpxA/lpxA, E. coli ΔwaaC/waaC, and E. coli ΔwaaY/waaY were monitored as described above.

TABLE 7.

Primers used to knock out lpxL and waaC by the CRISPR system

Primer	Sequencea	Description
lpxA-N20-F	TCCTAGGTATAATACTAGTGTGATGGTTGGCGGCTGCTCGTTTTAGAGCTAGAAATAGC	sgRNA of lpxA
waaC-N20-F	TCCTAGGTATAATACTAGTCCCAGGGATTAAGTTTGACTGTTTTAGAGCTAGAAATAGC	sgRNA of waaC
waaC-N20-F	TCCTAGGTATAATACTAGTACCCTTCATTTCGTACTTTTGTTTTAGAGCTAGAAATAGC	sgRNA of waaY
sgRNA-R	ACTAGTATTATACCTAGGACTGAGCTAGCT	Amplified plasmid pTargetF
lpxA-ssDNA	CGCAGTCCATCAGTTCTGCATCATTGGTGCGCACGTGATGGGCGCAGGACGTCCCTCCTTATGTCATTGCGCAGGGTAAC	Template for homologous recombination
waaC-ssDNA	CGTTGCCCGCACTCACTGATGCCCAGCAGGCAATCCCAGGTGGAAGAAGGGTTCGCACAGATTCCTTCCTGGCACGCTGC	Template for homologous recombination
waaY-ssDNA	CCAATAAATAAAAGTCATTGAGTGTATTTAACCCTTCATTTCGTACTGGGTTTGCTCAAAAAGGCGTTCGTAATAATCACCTTT	Template for homologous recombination

^aUnderlining indicates 20 bp of single guide RNA, which paired with complementary bases of target gene.

Assessment of synergy between OP4 and antibiotics.

The MICs of the mixtures of OP4 and antibiotics (Table 3) were determined using a checkerboard assay (52, 53). The fractional inhibitory concentration index (FICI) was calculated as dx/Dx + dy/Dy, where Dx and Dy are the MICs of individual drugs required to exert the same effect as the MICs dx and dy used in combination treatment. According to the European Committee for Antimicrobial Susceptibility Testing (EUCAST) (54), a synergistic effect is indicated when the FICI values are ≤0.5, whereas FICI values of >0.5 to ≤1, >1 to <2, and >2 are indicative of additive, indifferent, and antagonistic effects, respectively.

Detection of antibiotic accumulation in E. coli.

E. coli ATCC 35218 and the modified strains, namely, E. coli ΔlpxA, ΔwaaC, ΔwaaY, ΔlpxA/lpxA, ΔwaaC/waaC, and ΔwaaY/waaY (Table 5), were cultured in LB medium to mid-logarithmic phase. The cells were harvested, washed twice with PBS (50 mM, pH 7.2), and incubated with either antibiotics at half the MIC or OP4 (0.78 μg/mL) plus antibiotics (half the MIC) for 1 h at 37°C. The cells were then vortexed for 30 min after adding glass beads (diameter = 0.1 mm) and PBS (pH 2.0) to facilitate their lysis. The supernatant was passed through a solid-phase extraction column (Agilent Bond Elut C₁₈ SPE column), and the antibiotic was eluted with 5 mL methanol followed by 2.5 mL isopropyl alcohol. After drying with nitrogen, the antibiotics were dissolved in methanol. Finally, the concentration of antibiotics was monitored by liquid chromatography-mass spectrometry (LC-MS) (Triple Quad 5500; SCIEX, MA, USA).

Antibiotic resistance development experiments.

E. coli was cultured at half the MIC of antibiotics (250 μg/mL Ery and 0.03 μg/mL CIP) in LB medium for 24 h. Then, the new MICs of antibiotics were determined and the strain was cultured at half the new MICs of the antibiotics. In this way, the reciprocating cycle was cultivated to 30 generations (lasting 30 days). After adaptation, the bacteria were grown in LB medium without antibiotics for 15 generations (i.e., for 15 days) to observe the sustainability of the acquired resistance. The MIC was determined daily. The experiment was performed in triplicate.

Cytotoxicity assay.

Cell toxicity was detected as follows (55). Colon adenocarcinoma cells (Caco-2) and human liver carcinoma cells (HepG2) were cultured in Dulbecco’s modified Eagle’s medium (Sigma-Aldrich) and seeded at a density of 10⁴ cells/well in 100 μL culture medium. Subsequently, serially diluted solutions of PlnA1 (10 μL) were added to the cells and incubated for 24 h. After incubation, 50 μL MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide] (0.5 mg/mL; APExBIO Technology LLC, TX, USA) was added to the cells in each well, followed by incubation for an additional 4 h. The cells were harvested and resuspended in 150 μL dimethyl sulfoxide (Sigma-Aldrich), and the absorbance at 570 nm was detected with a spectrophotometer (Shimadzu Corporation, Kyoto, Japan).

Murine sepsis model.

Specific-pathogen-free BALB/c mice (8-week-old males, 22 to 25 g each) were obtained from the Comparative Medicine Centre of Yangzhou University, Jiangsu, People’s Republic of China [animal quality certificate no. 202111661; license no. SYXK(SU)2017-0007]. The mice were housed in groups of five animals per individually ventilated cage and maintained on a regular 12-h light-dark cycle [Laboratory of Animal Center, Nanjing Agricultural University, license no. SYXK(SU)2011-0036]. All animal care and experimentation were conducted in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care guidelines NRC2011 (https://www.aaalac.org/the-guide/) and were approved by the Laboratory Animal Welfare and Ethics Committee (Nanjing Agricultural University, license NJAU no. 20210517071).

The time frame of the murine sepsis study is illustrated in Fig. 8. Sepsis was established by intraperitoneal injection of 0.1 mL E. coli ATCC 35218 suspension (10⁷ CFU per mouse) on day 0. Solutions of Ery (10 mg/mL) and Ery+OP4 (1 mg/mL each) were prepared in 0.9% saline. At 2 h after the E. coli inoculation, the Ery treatment group of mice was injected with Ery (100 μg/g BW each), and the Ery+OP4 group was injected with Ery+OP4 (10 μg/g BW each), whereas saline was injected into the control group of infected mice. On days 1 and 7, half of the animals were anesthetized using tribromoethanol (200 μg/g BW). Blood was collected via cardiac puncture using a 2-mL vacuum blood collection tube (containing EDTA-K₂), and the plasma was collected via centrifugation at 5,000 × g for 5 min. In addition, 0.1-g samples of diverse organs, such as the heart, liver, spleen, lung, and kidney, were homogenized with 0.9 mL of 0.9% saline. Blood and organ homogenates were serially diluted in sterile 0.9% saline, plated on LB agar plates, and incubated at 37°C overnight. The bacterial colonies were then counted and used for several calculations. Additionally, levels of inflammation-related cytokines, including CRP, TNF-α, IL-1β, IL-6, and NF-κB, were quantified in the plasma using the respective ELISA kits (Sigma-Aldrich).

Statistical analysis.

The experiments in the present study employed a randomized design with four treatments with four repeats per treatment, and the average was calculated at the indicated times or at experimental endpoints. The data were analyzed using one-way analysis of variance, and post hoc comparisons were made using Tukey’s multiple-comparison tests in SPSS (IBM; version 17.0). Differences were considered statistically significant when P values were less than 0.05.

Data availability.

All data generated or analyzed during this study are included in the published article and the supplemental material.