Effects of Polyethyelene Glycol-Desferrioxamine:Gallium Conjugates on Pseudomonas aeruginosa Outer Membrane Permeability and Vancomycin Potentiation

Abstract

Pseudomonas aeruginosa exhibits a broad spectrum of intrinsic antibiotic resistance because of the limited permeability of its outer membrane. Given this situation, molecules that could make Gram-negative bacteria more permeable and more susceptible to large-scaffold Gram-positive antibiotics may be advantageous. Herein, we evaluate the antimicrobial activity of a series of targeted poly(ethylene glycol)-desferrioxamine/gallium (PEG-DG) conjugates that can improve the sensitivity of P. aeruginosa to the glycopeptide vancomycin (VAN). We observed that single-ended mPEG-DG and double-ended PEG-DG₂ conjugates characterized by PEG MW ≥2000 synergistically enhanced the sensitivity of VAN against P. aeruginosa reference strains PAO1 and ATCC 27853 and three clinically isolated carbapenem-resistant strains, but not Escherichia coli strain ATCC 25922. Although the exact mechanism of this phenomenon is currently under investigation, PEG-DG conjugates enhanced nitrocefin (NCF), hexidium iodide (HI), and VAN permeability only when PEG and DG were directly conjugated. The two most important physicochemical factors contributing to the synergistic activity observed with VAN relate to (1) the final concentration of DG ligands conjugated to the polymer and (2) the polymer length, wherein MW ≥2000 yielded a similar fractional inhibitory concentration.

Graphical Abstract

INTRODUCTION

The increasing prevalence of multidrug-resistant (MDR) Pseudomonas aeruginosa (P. aeruginosa), especially those strains expressing carbapenem-resistant phenotypes, has been labeled an urgent priority for the development of new antimicrobial strategies by the World Health Organization.¹ This rapidly growing threat to public health underlines a crucial need for new therapeutics to combat this pathogen.^2–4 Gram-negative bacteria are intrinsically resistant to many antibiotics because of the ineffective permeability of the outer membrane (OM) barrier. This membrane is an asymmetric bilayer of lipopolysaccharides (LPSs) and phospholipids, into which nonspecific porins and specific uptake channels are embedded. Small hydrophilic compounds with a molecular weight cutoff of around 600 Da (MW < 600) can diffuse through the OM via the water-filled porin channels.^5–7 However, many of the harmful agents, including antibiotics, are either hydrophobic or relatively large hydrophilic compounds which cannot effectively cross the OM. These large-scaffold antibiotics can, in some instances, diffuse through the LPS layer, but the process is inefficient and they are therefore considered ineffective against Gram-negative bacteria.^8–10 In Gram-negative bacteria, the accessibility to antibiotic target sites is restricted by limited OM permeability which plays a major role in multidrug resistance, especially in the case of P. aeruginosa which exhibits approximately 12-fold lower OM permeability relative to Escherichia coli.^10–13 Therefore, increasing the OM permeability of P. aeruginosa to antibiotics may be a tool to combat multidrug resistance.

Cationic steroid antibiotics and antimicrobial peptides can enhance OM permeability and improve the activity of large antibiotics against many Gram-negative species but are often limited by mammalian cell toxicity or low selectivity for P. aeruginosa.^14–17 Recently, our group has reported on the activity of Pluronic micelles as an alternative membrane permeabilization strategy and shown that the incorporation of a xenosiderophore-targeting group for ferrioxamine [i.e., desferrioxamine B (DFO):iron complex] receptors leads to pathogen-specific cell surface accumulation, OM permeabilization, and the potentiation of antibiotic activity against P. aeruginosa, although the roles of solution structures and individual polymer size, architecture, and hydrophobicity on this activity are yet to be elucidated.¹⁸

For further study, we present a small library of double-ended DFO:gallium (DG)-targeted polyethylene glycols (PEG-DG₂) and single-ended DG-targeted polyethylene glycol methyl ethers (mPEG-DG) to investigate the effect of polymer size and structure (dumbbell vs lollipop, respectively) on vancomycin (VAN) potentiation against P. aeruginosa. PEG in the range from 400 Da to 50 kDa is generally considered to be biologically inert at low concentrations and has many applications in the drug-delivery field because of its stealth behavior.¹⁹ Interestingly, PEGs have also been shown to interact with mammalian cell membranes, where their effects on membrane fluidity can lead to the inhibition of P-glycoprotein (P-gp)-mediated drug efflux from cancer cells.^20,21 In bacteria, PEG has been shown to prevent bacterial adhesion to surfaces but has not been characterized as having any membrane-level activity.²² By incorporating bacterial-targeting ligands such as DG at the end of PEG chains, we were able to significantly increase the accumulation of these polymers on the P. aeruginosa OM, which resulted in increased drug permeability and enhanced VAN sensitivity.

Targeting siderophore receptors may be particularly beneficial during an infection because P. aeruginosa must compete with the host for iron by upregulating the expression of OM receptors responsible for the uptake of siderophore:iron complexes, including xenosiderophores such as enterobactin and ferrioxamine (FO).^23,24 In this study, the FDA-approved iron-chelating drug DFO was complexed with gallium as a targeting ligand for P. aeruginosa.²⁵ DG complexes were conjugated to mPEGs and PEGs of varying molecular weights to investigate their effects on the bacterial OM and membrane permeability. VAN was selected as a model drug, as it has essentially no antipseudomonal activity alone but reaches its target site of action in the periplasm directly upon traversing the OM. For comparison, two additional commercial antibiotics erythromycin (ERY) and rifampicin (RIF) were also selected as model drugs because of their target cytoplasmic site of action and distinct mechanism of action. PEG-DG conjugates combined with VAN, ERY, and RIF were evaluated against two P. aeruginosa reference strains (ATCC 27853 and PAO1) and three clinically isolated carbapenem-resistant strains for sensitivity (Table S1),²⁶ with E. coli ATCC 25922 serving as a negative control organism because it has been reported not to recognize DG.²⁷

MATERIALS AND METHODS

Bacterial Strains, Media, and Chemicals.

P. aeruginosa reference strains used were ATCC 27853 and ATCC 15692 (PAO1). Clinically isolated MDR P. aeruginosa strains used (2638, 3072, and 24530) were generously provided by Dr. David Andes at the University of Wisconsin School of Medicine and Public Health.²⁶ The MDR strains selected for this study were chosen based on their resistance phenotypes, with MDR 2638 representing a pan-resistant organism, while MDR 3072 and MDR 24530 represent carbapenem-resistant organisms that are still susceptible to at least one clinically relevant antipseudomonal agent. The E. coli reference strain used was ATCC 25922. Antibiotic resistance profiles of strains used are summarized in Table S1.²⁶ P. aeruginosa and E. coli strains were grown in Mueller–Hinton Broth II (MHB) at 37 °C with agitation (250 rpm). DG complex, mPEG-DG, and PEG-DG₂ conjugates were synthesized and purified as described in the Supporting Information, and the synthetic DFO:Cr complex was prepared as reported in the literature.^28,29

Antimicrobial Efficacy.

Antimicrobial efficacy was assessed by the checkerboard broth microdilution assay as per the Clinical and Laboratory Standards Institute (CLSI) guidelines.³⁰ Organisms were grown overnight on cation-adjusted MHA plates and three morphologically distinct colonies were transferred to a culture tube containing cation-adjusted MHB. Liquid cultures were grown at 37 °C until the turbidity matched a 0.5 McFarland standard and diluted 1:20 in sterile water, followed by a 1:10 dilution when inoculating wells of a 96-well plate. Plates were then incubated at 37 °C for 18 h and the minimum inhibitory concentration (MIC) was read as the lowest concentration of the antimicrobial agent that had no visible growth and resulted in an OD₆₀₀ < 0.05 with respect to untreated growth controls. Synergistic activity was calculated using the following equation³¹

$$\text{FICI}=\frac{C_{\text{VAN,combination}}}{{\text{MIC}}_{\text{VAN,alone}}}+\frac{C_{\text{polymer-DG,combination}}}{{\text{MIC}}_{\text{polymer-DG,alone}}}$$

OM Binding Studies.

OM binding assays were conducted using fluorescein isothiocyanate (FITC)-labeled polymers using P. aeruginosa ATCC 27853 and E. coli ATCC 25922 as a negative control. Liquid cultures were grown in cation-adjusted MHB until the turbidity of the culture matched a 0.5 McFarland standard, after which aliquots were centrifuged and resuspended in 10 mM phosphate-buffered saline (PBS) (pH 7.4) containing either 256 μM mPEG5k-FITC or DG-PEG5k-FITC. Samples were incubated at 37 °C for 30 min and washed with 10 mM PBS. Control staining for the bacterial membrane was performed with FM 4-64FX; samples were fixed with 4% paraformaldehyde and directly imaged with a Zeiss LSM 710 confocal microscope (Figures 1, S9). Binding to the bacterial surface was also quantified by measuring the fluorescence emission intensity of each sample for FITC (ex/em = 495/520) and FM 4-64FX (ex/em = 545/665) using a fluorescent plate reader (SpectraMax Gemini, Molecular Devices).

Figure 1.

DG-PEG5k-FITC (MW 5000) exhibits increased binding to P. aeruginosa. (A) CLSM images of P. aeruginosa and E. coli incubated with mPEG-FITC (Green) (MW 5000) or DG-PEG-FITC (MW 5000) (Green) and the membrane stain FM 4-64FX (red). (B) Fluorescence intensity of DG-PEG-FITC bound to bacterial OM. (C) CLSM images of P. aeruginosa without and with pretreatment of DFO:Cr prior to incubation with DG-PEG-FITC (green). P. aeruginosa without DFO:Cr treatment shows more polymer accumulation, which suggests that ferrioxamine (DFO:Fe) OM receptors may be involved in the binding interaction. Scale bars represent 5 μm, ***p < 0.001 and ns = not significant.

Competition Assays.

Competitive OM binding assays for FITC-labeled polymers were performed using P. aeruginosa ATCC 27853. Liquid cultures were grown in cation-adjusted MHB until the turbidity of the culture matched a 0.5 McFarland standard, after which aliquots were centrifuged and resuspended in 10 mM PBS (pH 7.4) containing DFO:chromium (DFO:Cr) (5 mM). After 30 min incubation, 256 μM DG-PEG5k-FITC was added to the abovementioned solutions and incubated at 37 °C for another 30 min prior to imaging with a Zeiss LSM 710 confocal microscope.

OM Permeabilization with HI.

OM permeabilization of P. aeruginosa ATCC 27853 and E. coli ATCC 25922 was visualized using mPEG5k-DG as a representative conjugate in the presence of hexidium iodide (HI), a fluorescent stain for Gram-positive organisms that cannot diffuse through the OM of Gram-negative bacteria. Liquid cultures were grown in cation-adjusted MHB until the turbidity of the culture matched a 0.5 McFarland standard, after which aliquots were centrifuged, resuspended in 10 mM Tris (pH 7.4), and treated with either 256 μM mPEG5k-DG or 256 μM mPEG5k plus 256 μM free DG. Samples were incubated at 37 °C for 2 h, washed with 10 mM Tris, and stained with red fluorescent HI for 1 h followed by green fluorescent SYTO13 for 5 min. Samples were washed with 10 mM Tris and directly imaged with a Zeiss LSM 710 confocal microscope (Figure S10). Binding was also quantified by measuring the fluorescence emission intensity of each sample for SYTO13 (ex/em = 490/510) and HI (ex/em = 520/600) using a fluorescence plate reader (SpectraMax Gemini, Molecular Devices).

OM permeabilization of P. aeruginosa ATCC 27853 and E. coli ATCC 25922 was investigated for all PEG-DG conjugates synthesized using the nitrocefin (NCF) assay,³² in which NCF is known to undergo hydrolysis in the periplasm of Gram-negative bacteria. Liquid cultures were grown in cation-adjusted Mueller–Hinton broth until the turbidity matched that of a 2.0 McFarland standard, after which aliquots were centrifuged and resuspended in 10 mM PBS (pH 7.4). Samples were treated with either 256 μM mPEG2k-DG, 256 μM mPEG5k-DG, 256 μM mPEG10k-DG, 128 μM PEG2k-DG₂, 128 μM PEG4k-DG₂, 128 μM PEG10k-DG₂, or the same concentration of the unmodified polymer plus 256 μM free DG. NCF was added to each sample to a final concentration of 250 μg/mL and the UV–Vis absorbance of the hydrolyzed product (corresponds to its penetration into the periplasm) was measured at 485 nm over 4 h at 37 °C (Figure S11).

Cytocompatibility Studies.

The cytotoxicity of each PEG-DG conjugate was evaluated against HeLa cells in Dulbecco’s modified Eagle medium. Cells were seeded at 3000 cells/well in 96-well plates and cultured at 37 °C for 24 h before treatment. Following 48 h incubation, cell viability was assessed using a metabolism-based resazurin assay. The resazurin substrate was added to each well and incubated for 4 h before measuring the fluorescence (ex/em = 560/590) (SpectraMax Gemini, Molecular Devices) (Figure S12). Cell viability was calculated using the following equation

$$\text{cell viability}(\%)=100\times \frac{E_{\text{sample}}-E_{\text{blank}}}{E_{\text{control}}-E_{\text{blank}}}$$

Next, hemolytic activity of the polymer conjugates was assessed by measuring hemoglobin release from bovine red blood cells (RBCs). RBCs were washed with 10 mM PBS (pH 7.4), diluted to a final concentration of 2 × 10⁶ cells per well, and incubated with PEG-DG conjugates in a humidified, 5% CO₂ atmosphere at 37 °C for 2 h. Samples were then centrifuged to pellet intact RBCs and hemoglobin released into the supernatant was quantified by UV–Vis absorbance (Figure S13). Data were normalized with PBS representing 0% hemolysis and 1% Triton X-100 in PBS as a 100% hemolysis positive control (PC) using the following equation

$$\text{hemolysis}(\%)=100\times \frac{A_{\text{sample}}-A_{\text{blank}}}{A_{\text{control}}-A_{\text{blank}}}$$

Statistics and Software.

All experiments were conducted in triplicate and are presented as means ± standard deviations (SDs). Results were analyzed using Student’s t-test (GraphPad Prism version 7.00; GraphPad Software, La Jolla, CA, USA). Statistical significance was assessed at the 95% confidence level.

RESULTS

Synthesis of PEG-DG Conjugates.

DG complexes were first prepared from DFO and an aqueous slurry of Ga(OH)₃ by following established literature procedures.³³ The terminal hydroxy groups of mPEG (MW 750, 2000, 5000, and 10,000) and PEG (MW 1000, 2000, 4000, and 10,000) were first oxidized to carboxylic acids and then conjugated to the free amine of the DG complex via an amide bond to obtain PEG-DG conjugate solution unimers (Scheme S1, Figures S1 and S2) characterized by lollipop and dumbbell shapes, respectively, at 86–98% DG conjugation efficiency, based on atomic absorption spectroscopy and UV–Vis spectroscopy (Table S2).³⁴ In general, there was a trend toward an apparent size increase for the conjugates as the molecular weight of the PEG chain increased, based on dynamic light scattering results (Table S2).

Antimicrobial Activity of PEG-DG Conjugates Alone and in Combination with Commercial Antibiotics.

The antibacterial activity of each PEG-DG conjugate was evaluated alone and in combination with antibiotics against P. aeruginosa and E. coli (Table 1, Figures S3–S8). Synergistic activity was determined using the fractional inhibitory concentration (FICI). The susceptibilities of reference strains and MDR strains of P. aeruginosa and E. coli to the PEG-DG conjugate alone were greater than 1024 μM DG equivalents for all strains. VAN also demonstrated similar poor growth inhibition, with MIC values above 512 μg/mL for all P. aeruginosa strains and 128 μg/mL for E. coli ATCC 25922. Interestingly, all the mPEG-DG and PEG-DG₂ conjugates with molecular weights higher than 2000 Da were synergistic with VAN against all strains of P. aeruginosa, while the shorter mPEG750-DG and PEG1k-DG₂ were markedly less effective. Although the shortest conjugates had little activity against P. aeruginosa, PEGs of MW ≥2000 investigated here had a similar activity, suggesting that a threshold polymer length is needed to enhance OM permeability. Unmodified PEG plus free DG did not have any synergistic activity with VAN, and none of the PEG-DG conjugates displayed any synergistic activity with VAN against E. coli perhaps because of the lack of DG receptors. Overall, partial growth curves for VAN combined with PEG-DG conjugates showed a dose-dependent relationship (Figures S3–S8). Additionally, VAN activity was amplified in both reference and MDR strains of P. aeruginosa, giving strong evidence that the mechanism of OM permeabilization using PEG-DG constructs is not affected by several of the most clinically relevant resistance phenotypes. No synergy was observed between PEG-DG conjugates with ERY for all P. aeruginosa and E. coli strains. RIF and PEG-DG conjugates only showed synergistic activity against P. aeruginosa strain ATCC 27853 but no synergy against other strains tested was found (Figures S9–S14).

Table 1.

Antimicrobial Activity of PEG-DG Conjugates in Combination with VAN, ERY, and RIF against P. aeruginosa and E. coli

		P. aeruginosa					E. coli
drug	polymer-DG conjugate	ATCC 27853a	PAO1a	MDR 2638a	MDR 3072a	MDR 24530a	ATCC 25922a
VAN	none	>1024	>1024	512	>1024	1024	128
	mPEG750-DG	>256 (N/A)	>256 (N/A)	256 (0.75)	>256 (N/A)	>256 (N/A)	128 (1.25)
	mPEG2k-DG	256 (0.50)	128 (0.38)	64 (0.38)	128 (0.38)	128 (0.38)	128 (1.25)
	mPEG5k-DG	128 (0.38)	64 (0.31)	128 (0.50)	128 (0.50)	128 (0.38)	128 (1.25)
	PEG1k-DG2	>256 (N/A)	>256 (N/A)	256 (0.75)	>256 (N/A)	>256 (N/A)	128 (1.25)
	PEG2k-DG2	128 (0.38)	128 (0.38)	64 (0.38)	128 (0.38)	128 (0.38)	128 (1.25)
	PEG4k-DG2	128 (0.38)	64 (0.31)	64 (0.38)	128 (0.38)	128 (0.38)	128 (1.25)
	PEG10k-DG2	128 (0.38)	64 (0.31)	64 (0.38)	128 (0.38)	128 (0.38)	128 (1.25)
ERY	none	256	256	512	512	512	32
	PEG2k-DG2	128 (0.75)	128 (0.75)	512 (>1.00)	512 (>1.00)	512 (>1.00)	16 (0.75)
	PEG4k-DG2	256 (>1.00)	256 (>1.00)	512 (>1.00)	512 (>1.00)	512 (>1.00)	16 (0.63)
	PEG10k-DG2	256 (>1.00)	256 (>1.00)	256 (0.75)	256 (0.75)	512 (>1.00)	16 (0.63)
RIF	none	32	16	8	16	16	4
	PEG2k-DG2	8 (0.38)	8 (0.56)	4 (0.63)	8 (0.63)	8 (0.63)	4 (>1.00)
	PEG4k-DG2	8 (0.38)	8 (0.38)	4 (0.63)	8 (0.63)	8 (0.63)	4 (1.25)
	PEG10k-DG2	8 (0.38)	8 (0.63)	4 (0.63)	8 (0.63)	8 (0.63)	4 (>1.00)

^aInhibitory concentrations for antibiotics are given in μg/mL, followed by FICIs given in parentheses. The MIC of each PEG-DG conjugate alone or for free DG was greater than 1024 μM for all strains, indicating no activity. An FICI <0.5 is considered synergistic, 0.5 ≤ FICI ≤ 4 is considered nonantagonistic, and FICI >4 is considered antagonistic. Only PEG constructs of MW ≥ 2000 display synergy with VAN.

OM Binding Assay.

To investigate the binding of PEG-DG conjugates to the OM, representative PEG5k conjugates were labeled with FITC. The bacterial membrane was labeled with the red fluorophore FM 4-64FX. Confocal laser scanning microscopy (CLSM) visually confirmed that notably more green fluorescence (i.e., DG-PEG5k-FITC) accumulated on the cell surface of P. aeruginosa compared to the untargeted mPEG5k-FITC (Figures 1A,B and S15). E. coli had approximately the same degree of nonspecific binding when incubated with either mPEG5k-FITC or DG-PEG5k-FITC.

In order to confirm that PEG-DG binding activity was dependent on a siderophore-mediated iron transport system, P. aeruginosa dispersed in iron-deficient medium (PBS 10 mM) was first treated with chromium complexes of DFO (DFO–Cr) for 30 min as previously reported^28,29 followed by 256 μM DG-PEG5k-FITC. Because DFO:Cr complexes are structural analogues of DG, they are recognized by the same xenosiderophore receptors in the OM but are kinetically inert compared to DG and therefore inhibit the transport of the bound complex through the transporter (leads to saturation of OM receptors).²⁸ Indeed, results show poor accumulation of DG-PEG5k-FITC on the cell surface following treatment with DFO:Cr (Figure 1C).

OM Permeabilization in P. aeruginosa.

To visualize OM permeabilization in P. aeruginosa, CLSM was used to track the accumulation of hexidium iodide (HI), a dye with poor permeability across the OM of Gram-negative organisms.³⁵ Treatment of P. aeruginosa with mPEG5k-DG resulted in markedly more HI accumulation compared to cells treated with the unmodified mPEG5k plus free DG (Figures 2A,B, S16). E. coli did not display any appreciable differences in HI penetration upon treatment with either of the conjugates, further supporting that the selective binding of PEG5k-DG conjugates via OM receptors is necessary to permeabilize the membrane.

Figure 2.

PEG-DG conjugates MW ≥2000 increase the OM permeability of P. aeruginosa. (A) CLSM images of P. aeruginosa and E. coli stained with HI (red) after mPEG5k-DG (128 μM) and mPEG5k (128 μM) plus free DG (128 μM) treatment. OM permeabilization results in accumulation of HI. PC staining of the membrane was performed with SYTO13 (green). Scale bars represent 5 μm. (B) Fluorescence intensity of HI after bacteria treated with mPEG5k-DG and mPEG5k plus free DG. (C) OM permeability was evaluated by measuring NCF hydrolysis after P. aeruginosa and E. coli treated with PEG-DG₂ conjugates (128 or 256 μM) and unmodified PEGs (128 or 256 μM) plus free DG (256 μM). ***p < 0.001, ns = not significant.

OM permeability was also evaluated kinetically using the poorly permeable chromogenic probe nitrocefin (NCF). NCF is typically excluded by the outer cell membrane of Gram-negative bacteria and is a useful reporter for following the kinetics of OM permeability because it is known to undergo hydrolysis by β-lactamases present in the periplasm of P. aeruginosa, resulting in a red color that can be measured at 485 nm by UV–Vis (A485).³² P. aeruginosa cells incubated with PEG-DG conjugates (MW ≥ 2000) showed significantly more NCF hydrolysis over 4 h compared to bacterial cells treated with the unmodified PEGs plus free DG (Figures 2C, S17). The small amount of NCF hydrolysis observed for E. coli is likely due to its higher intrinsic OM permeability to the dye compared to P. aeruginosa.¹⁰

Cell Toxicity and Hemolysis.

The cytotoxicity of PEG-DG conjugates was evaluated with a metabolism-based resazurin assay using HeLa cells as a model human cell line. All the PEG-DG conjugates as well as free DG were markedly less toxic than free DFO, suggestive of a good safety profile against mammalian cells (Figures 3A, S18).

Figure 3.

Mammalian cell toxicity of representative polymer-DG conjugates. (A) No cytotoxicity observed at the highest tested concentrations of each PEG-DG conjugate after 48 h. (B) <1% hemolysis occurred in RBCs treated with the highest tested concentration of PEG-DG conjugates after 2 h incubation at 37 °C relative to PC. The highest tested concentrations of all PEG-DG conjugates, at which minimal toxicity was observed, are considerably higher than the concentrations tested to potentiate VAN activity, indicating pathogen-specific activity.

A hemolysis assay using bovine RBCs was used to evaluate cell membrane damage by measuring hemoglobin release after a 2 h incubation, relative to a 1% Triton X-100 total lysis PC. PEG-DG conjugates resulted in <1% hemolysis at the highest concentrations tested, suggesting low permeabilization to mammalian cell membranes (Figures 3B, S19).

DISCUSSION

Antibiotic resistance is a major public health threat around the world. Although new antibiotics are urgently needed and numerous efforts have been made to discover drugs with new mechanisms of action, the antibiotic discovery pipelines are still limited, particularly for Gram-negative bacteria. Given this situation, we need to find a new strategy to fight these emerging MDR bacteria. As the OM of Gram-negative bacteria is a robust permeability barrier that prevents many antibiotics from reaching their intracellular targets, molecules that could make Gram-negative bacteria more permeable and more susceptible to these antibiotics could prove extremely versatile to our arsenal of drugs in the fight against drug resistance. With this purpose in mind, we set out to identify molecules that could make Gram-negative bacteria more permeable and therefore more susceptible to large-scaffold antibiotics. In a previous study, we reported that the incorporation of a siderophore-targeting group such as DG on Pluronic micelles leads to enhanced pathogen-specific cell surface accumulation, OM permeabilization, and the potentiation of antibiotic activity against P. aeruginosa.¹⁸ Herein, to elucidate further the role of the DG complex from the potential permeabilization properties of the polymer, we investigated the effects of conjugating DG to hydrophilic PEGs of varying molecular weights in order to investigate their OM permeability properties on P. aeruginosa.

In general, PEGs are reported to display poor antibacterial activity, although PEG-400, PEG-1000, and PEG-15000 solutions at elevated concentrations have been reported to result in significant antibacterial activity against various pathogenic bacteria, including P. aeruginosa, in wound infections.^36–38 The antibacterial effect might be attributed to a combination of the lowering of water activity by the large concentration of PEGs at the membrane and because of the nonspecific interaction of the polymer.³⁹ In the concentration range investigated in our study, all PEG-DG conjugates possessed little antibacterial activity on their own against P. aeruginosa and E. coli strains (MIC > 1024 μM). Curiously, the ability of PEG-DG conjugates to potentiate VAN activity against P. aeruginosa displayed an unusual dependence on polymer length, wherein only PEG-DG conjugates of MW ≥2000 displayed synergistic activity with VAN (Table 1).

This suggests that a minimum critical polymer length may be necessary to successfully enhance the OM permeability of P. aeruginosa to enhance VAN penetration. Although evidence shows that some PEGs of specific molecular weights can interact with Caco-2 cell membranes and block P-gp,²¹ PEG2k has not been reported to interact with bacterial membranes in the micromolar concentration range, which was the range used in our study.⁴⁰ PEG2k alone did not measurably interact with bacterial cell membranes and no enhanced sensitivity to VAN was observed in the absence of DG conjugation for all the constructs in our studies. The permeability of VAN across the OM membrane appears to be highly dependent on both the selective interaction of PEG-DG with OM receptors and a minimum PEG length in order to sufficiently enhance VAN permeability across the OM. Furthermore, the number of DG complexes per polymer chain did not play a significant role in the effectiveness of OM permeability because similar FICIs were observed for the single-ended mPEG-DG and double-ended PEG-DG₂ conjugates. The two most important physicochemical factors contributing to the synergistic activity observed with VAN relate to (1) the final concentration of targeting ligands conjugated (i.e., DG) and (2) the polymer length wherein MW ≥2000. Differences in synergistic activity observed for each antibiotic are attributed to their different target sites. ERY and RIF bind to the 50S ribosomal subunit and RNA polymerase, respectively, both of which are cytoplasmic targets.^41,42 This requires ERY and RIF to diffuse across the inner membrane in order to have activity, so increased OM permeability alone does not allow direct access to the target site. VAN targets the cross-linking enzymes responsible for synthesis of the bacterial cell wall, which occurs in the periplasm of Gram-negative organisms.⁴³ Enhanced OM permeability allows VAN to directly reach its target site in P. aeruginosa, therefore resulting in highly synergistic activity with PEG-DG conjugates.

To further confirm that the interaction of PEG-DG conjugates MW ≥2000 to the OM is receptor-specific rather than nonspecific, the accumulation of DG-PEG5k-FITC on the OM of P. aeruginosa was visualized by CLSM and was observed to be much more elevated compared to mPEG5k-FITC (Figure 1A,B). Competition assays with DFO:Cr also confirm that if the receptors become saturated, PEG-DG conjugates poorly accumulate on the P. aeruginosa membrane (Figure 1C). In addition, the OM permeability of P. aeruginosa was observed to be much more susceptible to the effects of PEG-DG compared to E. coli, based on HI intensity being fivefold higher in P. aeruginosa after treatment with mPEG5k-DG compared to little differences between the constructs in E. coli (Figure 2A,B). Unconjugated PEG polymers plus free DG had no effect on OM permeability to VAN, which confirms that DG must be conjugated to the polymer to enhance bacteria sensitivity to VAN. Similar results were confirmed in permeability studies conducted using the NCF assay (Figures 2C, S17). Encouragingly, all the PEG-DG conjugates were furthermore markedly less toxic than free DFO, suggestive of a good safety profile against mammalian cells (Figure 3A) and caused <1% hemolysis at the highest concentration tested (1024 μM), suggesting low permeabilization to mammalian cell membranes (Figure 3B).

CONCLUSIONS

The results obtained in this study may provide new insights into the use of PEG-DG conjugates as OM permeability enhancers to improve penetration of poorly permeable antibiotics. mPEG-DG and PEG-DG₂ conjugates of MW ≥2000 possess no intrinsic antibacterial activity on their own (MIC > 1024 μM) but can accumulate on the cell surface of P. aeruginosa through specific interactions with target OM receptors of ferrioxamine and this property displayed synergy with VAN but not RIF or ERY. PEG-DG conjugates enhance VAN permeability only when PEG is directly conjugated to DG rather than due to the combined effects of PEG plus DG. Although the mechanistic details of this phenomenon are still unclear, the results of this study show that binding of PEG-DG conjugates MW ≥2000 to OM receptors of ferrioxamine leads to an enhanced permeability to VAN, HI, and NCF. The PEG-DG conjugates exhibited synergistic activity with VAN against reference strains and clinically relevant MDR strains of P. aeruginosa, pointing to their potential therapeutic use as sensitizers for selecting poorly permeable Gram-positive antibiotics in the fight against MDR Gram-negative bacteria. Finally, the highest tested concentrations of all PEG-DG conjugates, at which minimal toxicity was observed, are considerably higher than the concentrations tested to potentiate VAN activity, indicating pathogen-specific activity.