Intracellular Delivery and Antibacterial Activity of Gentamicin Encapsulated in pH-Sensitive Liposomes

Abstract

Cell membranes are relatively impermeable to the antibiotic gentamicin, a factor that, along with the toxicity of gentamicin, precludes its use against many important intracellular bacterial infections. Liposomal encapsulation of this drug was used in order to achieve intracellular antibiotic delivery and therefore increase the drug’s therapeutic activity against intracellular pathogens. Gentamicin encapsulation in several dipalmitoylphosphatidylcholine (DPPC) and pH-sensitive dioleoylphosphatidylethanolamine (DOPE)-based carrier systems was characterized. To systematically test the antibacterial efficacies of these formulations, a tissue culture assay system was developed wherein murine macrophage-like J774A.1 cells were infected with bacteria and were then treated with encapsulated drug. Of these formulations, DOPE–N-succinyl-DOPE and DOPE–N-glutaryl-DOPE (70:30;mol:mol) containing small amounts of polyethyleneglycol-ceramide showed appreciable antibacterial activities, killing greater than 75% of intracellular vacuole-resident wild-type Salmonella typhimurium compared to the level of killing of the control formulations. These formulations also efficiently eliminated intracellular infections caused by a recombinant hemolysin-expressing S. typhimurium strain and a Listeria monocytogenes strain, both of which escape the vacuole and reside in the cytoplasm. Control non-pH-sensitive liposomal formulations of gentamicin had poor antibacterial activities. A fluorescence resonance energy transfer assay indicated that the efficacious formulations undergo a pH-dependent lipid mixing and fusion event. Intracellular delivery of the fluorescent molecules encapsulated in these formulations was confirmed by confocal fluorescence microscopy and was shown to be dependent on endosomal acidification. This work shows that encapsulation of membrane-impermeative antibiotics in appropriately designed lipid-based delivery systems can enable their use in treating intracellular infections and details the development of a general assay for testing the intracellular delivery of encapsulated drug formulations.

Gentamicin is a polycationic aminoglycoside antibiotic with broad-spectrum antibacterial activity. Its use is indicated in several serious bacterial infections requiring hospitalization. The aminoglycosides are freely soluble in water, and after intravenous or intramuscular administration the majority of the drug remains in extracellular locations (29). Intracellular infections, caused by pathogens such as Salmonella, Listeria, and Mycobacterium species, constitute a challenge for classical antimicrobial therapies because of the requirement that antibiotics reach therapeutic levels at the intracellular site of infection. Thus, many antibiotics, such as gentamicin, that are active in vitro are often inactive against intracellular bacteria in vivo due to their poor penetration into cells at doses lower than the maximum tolerated dose or their inactivation by lysosomal enzymes (29). The development of new antibacterial formulations or carriers capable of intracellular delivery will improve therapy against infections that are presently difficult to treat.

Liposomal encapsulation of aminoglycoside antibiotics has been attempted by several groups as a means of altering the biodistribution of the drug and reducing its toxicity (3, 4, 6, 9, 12, 18, 24, 26, 33, 36, 41, 44). Egg phosphatidylcholine (EPC) encapsulation of gentamicin drastically increased the circulation time of the drug in rats compared to that of free drug and was active against Salmonella in a murine model (44). Fountain et al. (18) found enhanced antimicrobial activity with EPC-encapsulated gentamicin in the treatment of infections caused by the intracellular pathogen Brucella spp. both in vitro and in vivo. Other EPC and EPC-cholesterol based formulations have been used to treat infections caused by Salmonella dublin (12), and growth inhibition experiments have shown that liposomal gentamicin has enhanced efficacy over that of free gentamicin against Escherichia coli and Pseudomonas aeruginosa infections (41). Mycobacterium avium infections in human AIDS patients have also been treated with liposomal gentamicin with some success (36). In all but one case (9), gentamicin has been formulated in phosphatidylcholine (PC)-based liposomes with net neutral surface charges. Dees et al. (9) used neutral, anionic (dicetyl phosphate-containing), and cationic (stearylamine-containing) gentamicin-containing liposomes to treat Brucella abortus-infected mononuclear phagocytic cells isolated from bovine blood. Those investigators found that liposomal encapsulation enhanced the efficacy of the drug but that there was no difference between the formulations. None of these studies report on formulations specifically designed to facilitate the intracellular delivery of the antibiotic. Instead, the efficacies of these liposomal formulations were dependent on the ability of liposomal carriers to increase the lifetime of the encapsulated drug in the circulation and achieve passive accumulation at a site of infection. A liposomal formulation that can achieve cytoplasmic antibiotic delivery is likely to demonstrate significantly enhanced antibacterial activity beyond that conferred by either free drug or drug encapsulated in standard carriers.

The design of liposomal drug delivery systems that undergo controlled fusion with cellular or endosomal membranes has been the subject of intense effort (5, 27, 40, 42). The principle mechanism underlying many such “fusogenic” systems is the polymorphic phase behavior of dioleoylphosphatidylethanolamine (DOPE). While DOPE does not form a bilayer when dispersed in aqueous media, preferring to adopt the hexagonal-II phase, it can be stabilized into liposomes through the addition of lipids that favor a bilayer structure (8), including dioleoylphosphatidylcholine (DOPC) (48) or N-succinyl-DOPE (35). If the stabilizing lipid has a negatively charged head group with an appropriate pK_a, endosomal acidification can neutralize the lipid charge, reducing the bilayer-stabilizing effect. Bilayer destabilization in liposomes containing DOPE can result in both liposome fusion with adjacent membranes and content release (21, 22).

In this study, gentamicin was encapsulated in conventional, nonfusogenic liposomes as well as pH-sensitive fusogenic liposomal formulations. The latter were composed of DOPE stabilized by the addition of the pH-sensitive anionic lipids N-succinyl-DOPE or N-glutaryl-DOPE. It was anticipated that the abilities of these anionic lipids to stabilize DOPE would be pH dependent and that at endosomal pH values (30, 37) these carriers would be destabilized, release their encapsulated gentamicin into the endosomal lumen, and possibly disrupt the endosomal membrane, enhancing drug delivery to the cytoplasm. We have characterized the ability of these formulations to deliver gentamicin to Salmonella typhimurium and Listeria monocytogenes residing in cultured macrophages.

MATERIALS AND METHODS

Materials.

Commercially available lipids were obtained from Avanti Polar Lipids (Alabaster Ala.) or Northern Lipids (Vancouver, British Columbia, Canada). Polyethylene glycol (PEG)-C₂₀ (carbon chain length)-ceramide was manufactured at Inex Pharmaceuticals by Zhao Wang by modifications to a method described previously (47). The positively charged lipid dioleoyldimethyl ammonium chloride (DODAC) was synthesized by Michael Feng, also at Inex. Gentamicin sulfate was obtained from Sigma Chemical Company (St. Louis, Mo.). ³H- and ¹⁴C-labelled cholesteryl hexadecyl ether (CHE) as well as [³H]gentamicin were obtained from Amersham (Oakville, Ontario, Canada). The murine macrophage cell line J774A.1 and COS cells were obtained from the American Type Culture Collection (Rockville, Md.). Dulbecco’s modified Eagle medium (DMEM) and phosphate-buffered saline (PBS) plus Ca²⁺ and Mg²⁺ (PBS++) were obtained from Gibco (Burlington, Ontario, Canada). Fetal bovine serum (FBS) was obtained from Intergen (Purchase, N.Y.); calcein and 5-sulfofluorescein diacetate (5-SFDA) were obtained from Molecular Probes (Eugene, Oreg.). The detergent octaethylene glycol monododecyl ether (C₁₂E₈) was obtained from Fluka, Buchs, Switzerland. All other chemicals used were of reagent grade and were obtained from major suppliers.

Methods. (i) Lipid and gentamicin assays.

Lipid was quantified by either liquid scintillation counting (LSC) of samples containing a known quantity of [³H]- or [¹⁴C]CHE or by phosphate determination by the method of Fiske and Subbarow (16). Gentamicin levels were determined either by fluorescence polarization immunoassay performed by the clinical microbiology laboratory at Vancouver Hospital or by the addition of a known quantity of [³H]gentamicin followed by LSC.

(ii) Encapsulation of gentamicin in liposomes.

The lipid compositions of the various liposomes used in this study are provided in Table 1. Appropriate amounts of lipids were mixed in chloroform, and the solvent was evaporated by agitation under a nitrogen stream. Residual solvent was removed from the lipid mixture under high vacuum for at least 1 h. Dried lipid films (representing 25 mg of total lipid) were rehydrated by the addition of 1.0 ml of 32.25 to 64.5 mg of gentamicin (as the free base; corresponding to 50 to 100 mg of gentamicin per ml by dry weight) per ml in 20 mM HEPES–150 mM NaCl (pH 7.4) (HBS). Hydration was facilitated by extensive vortexing and five freeze-thaw cycles, and then this dispersion of multilamellar vesicles was converted to large unilamellar vesicles by 10 extrusions through two stacked 0.1-μm-pore-size filters (Poretics; AMD Manufacturing, Mississauga, Ontario, Canada) at 25 to 28°C with a Thermobarrel Extruder (Lipex Biomembranes, Vancouver, British Columbia, Canada) as described by Hope et al. (23). Liposome sizes were routinely determined by Gaussian analysis (intensity weighted; chi-squared, <0.4) of quasielastic light scattering with a NICOMP 7600 submicron particle sizer (NICOMP Systems, Santa Barbara, Calif.). All quantities of gentamicin subsequently stated in this paper refer to free base concentrations.

TABLE 1.

Summary of the physical characteristics of various gentamicin formulations

Formulation (molar ratio)	Final drug/lipid (wt/wt)	% Encapsulation	Mean liposome diam (nm [mean ± SD])
Neutral, DPPC-Chol (55:45)	0.075	2.9	134 ± 28
Cationic, DPPC-Chol-DODAC (30:40:30)	0.075	2.9	136 ± 31
Anionic
DOPE-DOPS-PEG (68:30:2)	0.073	2.8	205 ± 96
DOPE–N-succinyl-DOPE (70:30)	0.14	5.4	163 ± 63
DOPE–N-succinyl-DOPE–PEG (69.5:30:0.5)	0.14	5.4	142 ± 48
DOPE–N-succinyl-DOPE–PEG (69:30:1)	0.24	9.3	132 ± 53
DOPC–N-succinyl-DOPE (70:30)	0.16	6.2	102 ± 25
DOPC–N-succinyl-DOPE–PEG (69.5:30:0.5)	0.16	6.2	145 ± 60
DOPE–N-glutaryl-DOPE (70:30)	0.15	5.8	131 ± 50
DOPE–N-glutaryl-DOPE–PEG (69.5:30:0.5)	0.22	8.5	133 ± 40

Since gentamicin is a polycationic molecule, it was expected that a proportion of the drug would bind to the external surface of the anionic liposomes. The quantity of gentamicin bound to the outside surface of the liposomes was determined by incubating ³H-labelled gentamicin with empty liposomes for 30 min at 21°C at a drug-to-lipid ratio of 2:1 (wt/wt). The liposomes were then dialyzed overnight against HBS, the gentamicin content was measured by LSC, and the drug-to-lipid ratio was calculated and compared to that for an identical lipid film rehydrated as described above (i.e., which contained encapsulated plus externally bound drug). These preliminary studies (see Results) indicated the presence of significant quantities of antibiotic that bound to the external surfaces of anionic liposomes. Removal of gentamicin from the outside surfaces of these anionic liposomes was performed by adjusting the pH of the liposome suspension to 10 with a small quantity of glycine-buffered saline (1 M glycine, 150 mM NaCl [pH 10.0]). The liposomes were then passed through a 20- to 25-ml CM-Sepharose column equilibrated to pH 10.0 with 10 mM glycine–150 mM NaCl. The liposome fractions were collected and reequilibrated to pH 7.4 with 0.5 M HEPES–150 mM NaCl (pH 7.4). For DODAC- and rhodamine-containing formulations, purification was achieved by overnight dialysis against HBS followed by passage through a Sephadex G-50 column equilibrated to pH 10 as described above. The liposomes were subsequently readjusted to pH 7.4 as described above.

The ability of liposomal formulations to retain encapsulated antibiotic was determined by dialysis. Radiolabelled liposomes ([¹⁴C]CHE) that contained [³H]gentamicin were prepared, and externally bound drug was removed as described above. Samples were dialyzed (dialysis tubing with a molecular weight cutoff of 12,000 to 14,000) against at least 1,000 volumes of either HBS (pH 7.4) or DMEM plus 10% (vol/vol) FBS (pH 7.4) at 37°C. At various times 100-μl aliquots were removed and passed through 1-ml Sephadex G-50 spin columns (7) to remove the gentamicin that was external to the liposomes but inside the dialysis bag. Column eluates were analyzed by LSC, and the retention was calculated as the percent drug-to-lipid ratio relative to that at the initial time point.

(iii) Binding and uptake assay.

The mouse macrophage-like cell line J774A.1 was used to study the binding and uptake properties of the various liposome formulations. Cells were seeded at 5 × 10⁵ cells/well in 24-well plates in DMEM plus 10% FBS and were grown overnight. After 18 h, the cells were washed three times in PBS++, and then radiolabelled liposomes (200 μl) were added to a final lipid concentration of 1 μmol/ml. The cells were incubated with the liposomes for 2 h at either 4 or 37°C in 5% CO₂. After liposome binding and/or uptake, the cells were washed three times in PBS++ and were solubilized in 1% Triton X-100 and 0.1% sodium dodecyl sulfate (in PBS++) for 5 min, and then the cell-associated radioactivity of the lysate was determined by LSC. The cell-associated radioactivity observed at 4°C was due to surface binding of the liposomes, whereas the radioactivity observed at 37°C was taken to be the sum of binding and uptake. Hence, uptake of liposomes into J774A.1 macrophages was determined by subtraction of the data obtained at 4°C from the data obtained at 37°C.

(iv) Intracellular antibacterial efficacy assay.

Free gentamicin and various liposomal formulations were tested for their ability to kill intracellular pathogens infecting J774A.1 macrophages in a previously validated in vitro cell infection model (11, 45). The bacteria used in these experiments were either wild-type S. typhimurium SL1344, recombinant S. typhimurium SL1344 Hly⁺ expressing the E. coli hemolysin (51), or wild-type L. monocytogenes L028. S. typhimurium strains were cultured by standing them at 37°C in Luria-Bertani broth overnight, while L. monocytogenes was grown overnight in shaking cultures at 37°C in brain heart infusion (BHI) broth and was then subcultured for 2 h. All bacteria were stored on plates at 4°C or in frozen stock at −70°C. Before cell infection the bacterial suspensions were washed once in PBS++ to remove the accumulated hemolysin and were then adjusted to an optical density of 0.5 at 600 nm. J774A.1 cells plated at 5 × 10⁵ cells/well in 24-well plates were grown overnight at 37°C in 5% CO₂ in DMEM plus 10% FBS. For invasion of J774A.1 cells with Salmonella, the cells were infected with 5 × 10⁵ CFU/1 × 10⁶ cells for 15 to 20 min at 37°C in 5% CO₂. For the invasion of J774A.1 cells with L. monocytogenes, the cells were infected with 1.0 × 10⁶ CFU/1 × 10⁶ cells for 30 to 40 min at 37°C in 5% CO₂. After infection, the J774A.1 cells were washed three times with PBS++ and were then treated with gentamicin (100 μg/ml) for 1 h (conditions under which gentamicin is membrane impermeative and almost exclusively kills extracellular bacteria). The cells were again washed twice in PBS++ after the gentamicin treatment.

To test for the killing of intracellular bacteria, free gentamicin or liposomal gentamicin preparations were added at this stage, and liposomal uptake was allowed to proceed for a further 2 h under the same conditions. Free gentamicin was routinely used as a control to assess the advantages of encapsulated drug over free drug under conditions in which gentamicin is poorly permeative to cells. The cells were treated with gentamicin at a concentration of 150 μg of gentamicin base per ml. This corresponded to lipid doses of 0.68 to 2.0 mg/ml (0.8 to 2.5 μmol/ml), depending on the drug-to-lipid ratio (Table 1). Finally, the cells were washed three times with PBS++ and were then lysed in 1% Triton X-100–0.1% sodium dodecyl sulfate in PBS++. Dilutions of the lysates were grown overnight at 37°C on either Luria-Bertani plates (S. typhimurium) or tryptic soy agar plates (L. monocytogenes), and the numbers of CFU were counted.

(v) In vitro pH-mediated lipid mixing assays.

A description of the fluorescence assay used to monitor membrane mixing has been given previously (43). Briefly, liposomes encapsulating gentamicin were prepared as described above. The liposomes contained two fluorescent moieties that quench each other by resonance energy transfer when in close proximity. In the presence of excess unlabelled liposomes, the lipid mixing resulting from membrane lipid mixing dilutes the fluorescent probes in the membrane, causing fluorescence dequenching. Donor vesicles were prepared as described above, but they also contained 0.5 mol% N-4-nitrobenzo-2-oxa-1,3-diazole phosphatidylethanolamine (NBD-PE) and N-(lissamine rhodamine B sulfonyl)dioleoylphosphatidylethanolamine (Rh-PE). Acceptor vesicles were of the same composition as the donor vesicles but without the fluorescent molecules and with no encapsulated gentamicin. Donor liposomes in 10 mM HBS (pH 7.4) were mixed in a quartz cuvette (with 200 μl of FBS added when required) with acceptor liposomes (in 10 mM HBS [pH 7.4]) and brought to a final volume of 1,950 μl with 150 mM NaCl. When required, CaCl₂ was added to the mixture to a final concentration of 20 mM. The added volumes of donor and acceptor liposomes were adjusted to achieve final lipid concentrations of 0.1 and 1.0 mM, respectively. The cuvette was placed in an SLM Aminco-Bowman series II luminescence spectroscope with a stirring mechanism and was maintained at 37°C. After a 2-min stabilization period, the fluorescence emission at 535 nm (NBD fluorescence) was measured at an excitation wavelength of 445 nm, with 4-nm slit widths and the initial fluorescence set between 50 and 70% full scale. After a stable baseline was verified, 50 μl of acid solution (2.5% [vol/vol] acetic acid plus 250 mM citrate-buffered saline [250 mM citrate, 150 mM NaCl; pH 4.5] or dilutions thereof) was added directly to the cuvette. Fluorescence was measured continuously until a plateau was achieved, and then 100% dequenching of the NBD was determined by the addition of 50 μl of 0.2 M C₁₂E₈ detergent. The final pHs of the solutions were measured with a pH meter. The data were processed with software provided by Aminco-Bowman. Percent NBD dequenching, equivalent to percent fusion, was determined by the following formula: percent dequenching = (acid-induced fluorescence − initial fluorescence)/(C₁₂E₈-induced fluorescence − initial fluorescence) × 100.

(vi) Microscopy of J774A.1 cells incubated with liposomal formulations.

J774A.1 cells (10⁵) were incubated with formulations composed of either DOPE–N-succinyl-DOPE–PEG (69.5:30:0.5) or DOPC–N-succinyl-DOPE–PEG (69.5:30:0.5) encapsulating 5-sulfluorescein diacetate (5-SFDA) (initial dye:lipid ratio = 1:5) (10, 38, 39). The cells were incubated with the liposomes at a final lipid concentration of 250 nM for 2 h at 37°C. At the same time, the cells were incubated with Texas red ovalbumin (500 μg/ml), which fluorescently labels the endocytic pathway. The labelling was performed in the presence and absence of bafilomycin (1 μM), an inhibitor of endosomal acidification (50). The cells were then washed with PBS and incubated for a further 2 h in fresh growth medium. The cells were then examined by confocal microscopy (Bio-Rad model MRC600 with Bio-Rad COMOS software). A krypton/argon laser was used with a green filter (λ_Excitation, 488 nm; λ_Emission, 515 nm) to determine the fluorescein fluorescence. Texas red ovalbumin fluorescence was determined with a red filter (λ_Excitation, 568 nm; λ_Emission, 585 nm). The objective lens was a ×63 oil immersion lens. Sections (0.2 μm) were projected and visualized with NIH image, version 1.59, and the final montage was created with Adobe Photoshop, version 3.04.

RESULTS

Formulation and characterization of liposomal gentamicin.

The characteristics of the liposomal gentamicin formulations are summarized in Table 1. The final size of these formulations was typically about 135 nm (Table 1). For liposomes of this size at an initial concentration of 25 mg/ml and a trap volume of 1.1 μl/μmol of lipid, the theoretical maximum encapsulation efficiency for gentamicin is 3.9%. Net neutral liposomal formulations of gentamicin (dipalmitoylphosphatidylcholine [DPPC]-cholesterol [Chol] [55:45]) and cationic formulations (DPPC-Chol-DODAC) not expected to have an electrostatic association of the drug with the liposome had encapsulation efficiencies lower than this theoretical value (Table 1). For these neutral and cationic formulations, the starting drug/lipid ratio was 2.58 (64.5 mg gentamicin per ml and 25 mg of lipid per ml), and the final drug/lipid ratios were in the range of 0.075 (wt/wt) (Table 1). In contrast, the encapsulation of gentamicin in anionic liposomes was complicated by the electrostatic association of the polycationic drug with the anionic liposome surface. This interaction caused extensive liposome aggregation and poor recovery of the encapsulated drug in preliminary formulations (data not shown). However, aggregation could be ameliorated by the presence of small quantities of PEG-ceramide (≥0.25 mol%). Consequently, we routinely formulated gentamicin in anionic liposomes with 0.5 mol% PEG-ceramide (Table 1). Anionic formulations containing N-succinyl-DOPE and N-glutaryl-DOPE had drug-to-lipid ratios of between 0.138 and 0.241, representing encapsulation efficiencies of 5.5 to 9.3%. These values were higher than the maximum expected ratio of 0.10 (encapsulation efficiency, 3.9%) for passive encapsulation, suggesting a significant electrostatic association of gentamicin with these liposomes. Interestingly, formulations containing the anionic lipid dioleoylphosphatidylserine (DOPS) had encapsulation efficiencies lower than the theoretical values, suggesting that the exact chemical and/or physical disposition of the anionic lipid is important in the interaction with the polycationic drug.

Gentamicin is known to have an affinity for lipid membranes in general and anionic membranes in particular (1, 2, 17, 25, 28, 31, 32, 46). The external gentamicin bound to N-succinyl-DOPE-containing liposomes was quantified in experiments in which gentamicin was added to empty anionic liposomes and incubated for 30 min and then unbound drug was removed by chromatography on Sephadex G-50 columns at various pHs. These preliminary experiments indicated that approximately 25 to 33% of total gentamicin was associated with the outer surfaces of the liposomes (data not shown). Removal of this external gentamicin was achieved by passage of the liposome sample through Sephadex G-50 spin columns equilibrated at either pH 10 or pH 11 (data not shown). Both treatments removed the majority (75%) of the external gentamicin. CM-Sepharose columns equilibrated at pH 10 were then routinely used for this purpose prior to further in vitro testing (see Materials and Methods). Consequently, ≤8% of the final liposome-associated drug was exposed on the exterior surfaces of these liposomal carriers.

The retention of gentamicin was determined for two formulations under different buffer conditions. HBS (pH 7.4) was used because it is the standard storage buffer for these liposomes. DMEM plus 10% FBS was used to simulate the conditions under which the efficacy assay was performed. Neutral DPPC-Chol (55:45) liposomes showed very little leakage under either of the conditions. For this formulation, after 22 h of incubation in HBS at 37°C, the drug-to-lipid ratio was 89% of the initial value. In DMEM plus 10% FBS, the drug-to-lipid ratio at 24 h was 95% of the initial value. The DOPE–N-succinyl-DOPE formulation also showed negligible leakage in HBS (the drug-to-lipid ratio at 22 h was 96% of the initial value) or in DMEM plus 10% FBS (the drug-to-lipid ratio at 24 h was 93% of the initial value). In these experiments, in the case of the DOPE–N-succinyl-DOPE formulation, gentamicin could leak from the vesicles and bind to the outer leaflet. This could result in a maximum unobservable leakage of 25% (see above). Leakage exceeding 25% was not observed; therefore, the majority of the drug remained encapsulated in these liposomal carriers over 24 h at 37°C in DMEM plus 10% FBS. In contrast, during intracellular killing assays (see next section), the liposomes were exposed to DMEM plus 10% FBS for only 2 h at 37°C. It is concluded that under the conditions used for the intracellular killing assays, the drug remained encapsulated in the liposomal carriers.

Uptake of liposomal gentamicin by J774A.1 cells and intracellular killing.

The initial steps in liposomal drug delivery to cells in culture are carrier binding followed by endocytosis. Therefore, binding and uptake into J774A.1 cells were assessed. Figure 1A shows the characteristics of uptake by J774A.1 cells of several formulations at 1 μmol of lipid/ml. The formulation taken up by J774A.1 cells in the greatest quantities was the cationic DPPC-Chol-DODAC (30:40:30) formulation. In contrast, the anionic formulation, DPPC-Chol-DOPS (30:40:30) was poorly endocytosed by the cells. This result is likely due to the electrostatic attraction between the net negative charge of the cell surface and the cationic liposomes. The binding and uptake of neutrally charged DPPC-Chol liposomes by J774A.1 cells were also very low (data not shown). However, for the DOPE–N-succinyl-DOPE formulations, although they are negatively charged, uptake was enhanced over that for the anionic DOPS-containing formulation but uptake was less than that for the cationic formulation. The addition of 0.5 to 1.0 mol% PEG-C₂₀-ceramide to these formulations did not significantly reduce the level of uptake of these liposomes by J774A.1 cells (Fig. 1A). At proportions between 2 and 5 mol%, PEG-C₂₀-ceramide reduced the total uptake of DOPE–N-succinyl-DOPE formulations by J774A.1 cells (data not shown). Figure 1B shows that when cells are incubated with equivalent concentrations of either free gentamicin or gentamicin formulated in DOPE–N-succinyl-DOPE–PEG (69.5:30:0.5), the liposomal formulation delivers 21.5 times more antibiotic to the interior of the cell than simple diffusion or pinocytosis of free drug across the cell membrane. This result, combined with the uptake of the liposomal carriers into the J774A.1 cells (Fig. 1A) and the excellent retention of the drug within the carriers in the presence of DMEM (see above), demonstrates that the liposomes containing the encapsulated drug were being endocytosed by the cells. That is, the antibiotic did not leak from the formulation while it was outside the cells and then undergo greater fluid-phase uptake or passive diffusion into the cells.

FIG. 1.

Uptake and binding of vesicles and gentamicin to J774A.1 cells. (A) Various formulations of gentamicin were incubated with J774A.1 cells (5 × 10⁵) for 2 h at either 37 or 4°C at 1 μmol of lipid/ml. Uptake for each formulation was quantitated by subtraction of the result obtained at 4°C (binding only) from the result obtained at 37°C (uptake plus binding). The full bar is the sum of uptake plus binding. The uptake component is shown as unhatched bars; the binding component is shown as hatched bars. (B) J774A.1 cells were incubated with 150 μg of gentamicin per ml (doped with [³H]gentamicin) as either free drug or drug encapsulated in DOPE–N-succinyl-DOPE–PEG (69.5:30:0.5) vesicles. Excess drug was washed off, the cells were collected, and the amount of drug associated with the cells was determined by scintillation counting. Uptake for each sample was quantitated by subtraction of the result obtained at 4°C (binding only) from the result obtained at 37°C (uptake plus binding).

The biological activity of the drug delivery systems was evaluated by determining their efficacies against intracellular infections. A tissue culture assay was developed in which cells were first infected with intracellularly residing bacteria and were then exposed to external gentamicin formulations. The abilities of the different formulations to kill intracellular infections with wild-type S. typhimurium SL1344 (phagosome resident) (13–15, 19, 20), recombinant S. typhimurium SL1344 expressing the E. coli hemolysin (cytoplasm resident) (51), and L. monocytogenes (cytoplasm resident) (39) are presented in Fig. 2 to 4, respectively. Neutral (DPPC-Chol [55:45]) and anionic (DPPC-Chol-DOPS [30:40:30]) vesicles did not show antibacterial activity in comparison to that of a no-treatment control (data not shown). Cationic DODAC-containing vesicles, for which the level of uptake by J774A.1 cells was the highest (Fig. 1), showed antibacterial activity that was less than or comparable to that of free drug against wild-type S. typhimurium (Fig. 2) or hemolysin-expressing S. typhimurium and L. monocytogenes (Fig. 3 and 4), respectively. In contrast, gentamicin formulations composed of DOPE–N-succinyl-DOPE (with or without PEG-ceramide) killed 76% of intracellular wild-type S. typhimurium (Fig. 2). Substitution of DOPC for DOPE in these formulations reduced the antibiotic activity to that observed for free gentamicin (Fig. 2 to 4).

FIG. 2.

Killing of intracellular wild-type S. typhimurium by encapsulated gentamicin. J774A.1 cells infected with bacteria were incubated with free or encapsulated gentamicin formulations at a drug dose of 150 μg/ml for 2 h. The cells were lysed and the lysate was plated on growth medium overnight. The bacterial colonies were counted, and the result was expressed as percent killing compared to that for untreated control cells. Representative assay results are presented here. Each result represents the mean of triplicate assays performed together. Error bars indicate standard deviations. Asterisks represent values found to be significantly different (confidence level, t = 0.05) from that for free gentamicin by statistical analysis (two-tailed Student t test).

FIG. 4.

Killing of intracellular L. monocytogenes by encapsulated gentamicin. J774A.1 cells were treated as described in the legend to Fig. 2. ND, not determined. The data presented here are as described in the legend to Fig. 2. Double asterisks represent values found to be significantly different (confidence level, t = 0.02) from that for free gentamicin by statistical analysis (two-tailed Student t test).

FIG. 3.

Killing of intracellular recombinant hemolysin-expressing S. typhimurium by encapsulated gentamicin. J774A.1 cells were treated as described in the legend to Fig. 2. The data presented here are as described in the legend to Fig. 2.

Wild-type S. typhimurium resides within the phagosomes of infected macrophages. To evaluate the influence of the subcellular location of the target bacteria on efficacy, invasion assays were also performed with bacteria residing in the cytosol: a recombinant S. typhimurium strain that expresses an E. coli hemolysin and L. monocytogenes. The recombinant S. typhimurium strain escapes the phagosome by hemolysin-mediated lysis of the phagosomal membrane and then survives free in the cell cytosol (51), whereas L. monocytogenes escapes the phagosome by endogenous listeriolysin activity (39). Gentamicin-containing vesicles composed of DOPE–N-succinyl-DOPE (with or without PEG-ceramide) were the most potent at killing cytosol-residing intracellular bacteria compared to the other formulations or free gentamicin (Fig. 3 and 4). Moreover, substitution of DOPC for DOPE resulted in the same level of drug uptake by the cells (Fig. 1) but significantly reduced the bactericidal effect.

The presence of small proportions (0.5 mol%) of PEG-ceramide in these vesicles had no influence on the liposome uptake by the cells (Fig. 1) or their antibacterial activities (Fig. 2 to 4) but dramatically facilitated the formulation process. Substitution of N-glutaryl-DOPE for N-succinyl-DOPE also produced formulations with very good antibacterial activities (Fig. 2 to 4). This activity was specific to anionic formulations containing only N-succinyl-DOPE or N-glutaryl-DOPE; substitution of these lipids for the anionic phospholipid DOPS produced vesicles with limited killing activity against recombinant S. typhimurium. When infected J774A.1 cells were incubated under identical conditions with empty DOPE–N-succinyl-DOPE–PEG liposomes mixed with external (unencapsulated) gentamicin to equivalent lipid and gentamicin levels, no antibacterial activity above that of free gentamicin was observed (data not shown).

In order to assess the relative effectiveness of the formulations in terms of the release of bioavailable gentamicin intracellularly after uptake, the bactericidal data presented above were treated by normalization to both free gentamicin activity and the amount of lipid taken up by the cells. This treatment leads to a measure of, first, the contribution of encapsulation to bactericidal activity (specifically, the percent killing by free gentamicin was subtracted from that for the test sample) and, second, the relative intracellular release of gentamicin by each formulation at equivalent uptake values (division of the encapsulated killing value by the level of liposome uptake by the macrophages). A comparison of the efficacies exhibited by the different formulations, after correcting for levels of uptake by the cells, is summarized in Fig. 5. The cationic formulation (DPPC-Chol-DODAC) had a very low level of killing activity when its activity was corrected for its high level of uptake by the cells (Fig. 1). Against wild-type S. typhimurium, this activity was less than that of free drug (Fig. 5), suggesting that this carrier does not release gentamicin but effectively sequesters the drug and reduces its bioavailability. In contrast, the anionic formulations comprising DOPE and N-succinyl-DOPE had killing activities that were four- to fivefold higher than that observed for either the DPPC-Chol-DODAC formulations or the DOPC–N-succinyl-DOPE–PEG-ceramide liposomes (Fig. 5). Therefore, per liposome taken into the cell, the most efficacious formulations were those containing DOPE, PEG-ceramide, and N-succinyl-DOPE.

FIG. 5.

Summary of the antibacterial activities of various gentamicin formulations. Corrected killing values were obtained by subtraction of the mean killing observed for free gentamicin from that for the encapsulated gentamicin and dividing by the nanomoles of lipid taken up per 10⁶ cells (Fig. 1). The data presented here are for wild-type S. typhimurium (hatched bars), hemolysin-expressing S. typhimurium (cross-hatched bars), and L. monocytogenes (open bars).

Mechanism of intracellular antibiotic delivery.

The DOPE– N-succinyl-DOPE–PEG-ceramide and DOPE–N-glutaryl-DOPE–PEG-ceramide formulations described above were designed to destabilize at endosomal pH values and release the encapsulated drug into the endosomal, phagosomal, and/or cytoplasmic compartments. The significant reduction in antibacterial efficacy observed when DOPE is replaced with the nonfusogenic lipid DOPC (Fig. 2 to 5) suggests that enhanced gentamicin delivery may occur following phosphatidylethanolamine (PE)-induced perturbation of the endosomal membrane at reduced pH. To further characterize this, the ability of the formulations to undergo lipid mixing was determined as a function of pH by an NBD-PE and Rh-PE fluorescence assay. Figure 6A shows the kinetics of the fluorescence dequenching (due to lipid mixing and probe dilution) for formulations at a pH of 4.86 to 4.87. Both N-succinyl-DOPE-containing formulations were found to undergo significant lipid mixing under acidic conditions in the presence of 10% serum in contrast to the mixing of the DPPC-Chol control. However, the efficacious (DOPE-based) formulation was found to exhibit a much greater rate and extent of lipid mixing than the DOPC-based control formulation. Figure 6B shows the extent of fluorescence dequenching at 1 min over the pH range from 4.6 to 5.2, the pH range of the late endosome or lysosome (30, 37). The DOPE-based formulation showed significantly greater activity than the DOPC-based formulation over this pH range. DPPC-Chol liposomes showed no fusion under these conditions (data not shown). In serum-free medium to which 20 mM Ca²⁺ was added (Fig. 6C), significant lipid mixing was observed with the DOPE-containing formulation over a similar pH range at 1 min, whereas virtually no dequenching was observed for the DOPC-containing formulation.

FIG. 6.

Lipid mixing of vesicles containing gentamicin. The lipid mixing activities of gentamicin-containing DOPE–N-succinyl-DOPE–PEG (69.5:30:0.5) (open squares), DOPC–N-succinyl-DOPE–PEG (69.5:30:0.5) (closed circles), and DPPC-Chol (55:45) (open triangles) were monitored by a resonance energy transfer fluorescence dequenching assay as described in the text. (A) Time course of mixing at pH 4.86 (for the DOPC-containing formulation) and 4.87 (for the DOPE- and DPPC-containing formulations). (B) pH dependence of lipid mixing after 1 min of incubation. (C) pH dependence of lipid mixing after 1 min in serum-free medium containing 20 mM Ca²⁺.

Microscopy of liposomal interaction with J774A.1 cells.

Photomicrographs of the formulation with maximal antibiotic activity, along with those of the DOPC analog control, are presented in Fig. 7. For these photographs, 5-SFDA was incorporated into the liposomes. The cells were colabelled with Texas red ovalbumin, which marks the endocytic pathway of the cell and in this experiment serves to illustrate the number of cells in the image field. Figure 7A shows the Texas red ovalbumin fluorescence for J774A.1 cells treated with 5-SFDA-containing DOPE–N-succinyl-DOPE–PEG liposomes. Figure 7B shows the intense fluorescence from 5-SFDA under these conditions, indicating that the marker molecule has been released from the carrier and has cleaved to the fluorescent product by cellular enzymes. In contrast, cells incubated with the DOPC-based control containing 5-SFDA show very little fluorescein fluorescence (Fig. 7F), even though there were an equivalent number of cells in the image field (Fig. 7E) and equivalent amounts of lipid and label were added. This indicates that release from the DOPC-based carrier has not occurred. When the same experiment was performed in the presence of bafilomycin, which inhibits endosome acidification, the 5-SFDA fluorescence of the cells incubated with the DOPE-based formulation was virtually eliminated (Fig. 7D). This shows that the release of this marker from this formulation is dependent upon destabilization of the carrier caused by acidification of the endosome. This experiment was also performed with formulations in which N-glutaryl-DOPE was substituted for N-succinyl-DOPE in both the DOPE- and DOPC-based formulations. The formulations behaved very similarly to their N-succinyl-DOPE-containing counterparts (data not shown).

FIG. 7.

Microscopy of drug delivery formulations. J774A.1 cells were incubated with DOPE–N-succinyl-DOPE (70:30) or DOPC–N-succinyl-DOPE (70:30) vesicles containing 5-SFDA. At the same time, the cells were coincubated with Texas red ovalbumin. The figure shows the fluorescence confocal images that were obtained. Gain (intensity) and black levels were adjusted with the DOPE–N-succinyl-DOPE–PEG formulation and were kept the same for all samples tested. (A) DOPE-containing formulation, Texas red ovalbumin fluorescence; (B) DOPE-containing formulation, 5-SFDA fluorescence; (C and D) as for panels A and B, respectively, but in the presence of bafilomycin; (E) DOPC-containing formulation, Texas red ovalbumin fluorescence; (F) DOPC-containing formulation, 5-SFDA fluorescence; (G and H) as for panels E and F, respectively, but in the presence of bafilomycin.

DISCUSSION

In this paper we report on the development of several liposomal formulations that achieve the intracellular delivery of the membrane-impermeative antibiotic gentamicin. We have shown that formulations capable of delivering gentamicin intracellularly have dramatically increased antibacterial activity over that of free gentamicin against intracellular pathogens such S. typhimurium, the agent of typhoid fever and salmonellosis, and L. monocytogenes, which is responsible for meningitis, septicemia, etc.

Gentamicin was loaded into lipid vesicles by a passive encapsulation process that proceeds with a low efficiency (3.9% for vesicles of this size and starting concentration). The efficiency of gentamicin encapsulation in anionic delivery systems was increased due to the electrostatic interaction of the polycationic drug with anionic lipid head groups (Table 1). However, under these conditions gentamicin also caused extensive aggregation of the vesicles. This could be ameliorated by incorporating small quantities (0.5 mol%) of PEG-ceramide into the lipid bilayer without affecting either the binding of carrier to the cells (Fig. 1) or antibacterial efficacy (Fig. 2 to 5). Mingeot-Leclerq et al. (31, 32) have reported that gentamicin binds to anionic lipids and that the interaction is sufficient to restrict the mobility of lipid head groups. We also found significant levels of drug associated with the external surfaces of our anionic formulations. A process was developed to remove the majority of this electrostatically bound drug, and control experiments demonstrated that the low levels of residual gentamicin at the vesicle surface had negligible antibacterial activity (data not shown).

The most efficacious gentamicin formulations described here are those in which the drug has been encapsulated in anionic vesicles containing DOPE and either N-succinyl-DOPE or N-glutaryl-DOPE (Fig. 2 to 5). Comparable formulations containing the anionic lipid DOPS showed antibacterial activity no greater than that of free gentamicin (data not shown). Even though cationic drug carriers were efficiently endocytosed by cells, the poor bactericidal activities of these formulations compared to that of free drug (Fig. 2 to 4) suggests that the majority of the drug was not bioavailable. It is concluded that the cationic and DOPS-based formulations of gentamicin were not effective at releasing the antibiotic from the endosomal compartment. Drug encapsulated in DPPC-Chol vesicles was also ineffective at killing intracellular pathogens (data not shown), but these neutral drug carriers were poorly endocytosed by the J774A.1 macrophages. Overall, these results show that good uptake into the target cells only is insufficient for improving the activity of liposomal drug over that of free drug. The carrier system must also enhance the bioavailability of intracellular gentamicin.

The ability of DOPE–N-succinyl-DOPE and DOPE–N-glutaryl-DOPE formulations to release bioavailable drug into the endosomal compartment is consistent with our observations concerning the intracellular delivery of fluorescent probes by the same vesicles. Preparations encapsulating quenched calcein or 5-SFDA showed by confocal microscopy (Fig. 7) enhanced release of probes compared to that for other formulations. Interestingly, these formulations also exhibited their greatest activity against the hemolysin-expressing S. typhimurium and L. monocytogenes strains residing in the cytoplasm compared to that against the wild-type S. typhimurium strain residing in the vacuole (Fig. 5). This suggests that the bacteria resident in the cytoplasm were exposed to drug released from the endosomes. The presence of a vacuolar membrane would be expected to act as a permeation barrier to cytoplasmic drug, reducing the dose intensity experienced by intravacuolar wild-type S. typhimurium.

Several lines of evidence suggest that the DOPE–N-succinyl-DOPE and DOPE–N-glutaryl-DOPE formulations achieve cytoplasmic delivery of gentamicin by perturbing or fusing with the endosomal membrane at reduced pHs. First, these anionic vesicles show significantly higher levels of pH-dependent lipid mixing (a standard assay for monitoring the fusion of model membrane systems) than vesicles that did not enhance the efficacy of gentamicin above that of free-drug controls (Fig. 6 and Fig. 2 to 5). In particular, substitution of DOPC for DOPE did not affect endocytosis but significantly reduced the level of bactericidal activity as well as the level of lipid mixing. Second, intracellular fluorescence measured by confocal microscopy (Fig. 7) after delivery of encapsulated calcein or 5-SFDA was greater with DOPE–N-succinyl-DOPE- and DOPE–N-glutaryl-DOPE-containing systems than with DOPC-containing systems. Finally, the increase in fluorescence following the intracellular delivery of 5-SFDA from either DOPE–N-succinyl-DOPE–PEG (Fig. 7) or DOPE–N-glutaryl-DOPE–PEG (data not shown) formulations was greatly reduced in the presence of bafilomycin, an inhibitor of endosomal acidification. This result, observed in J774A.1 cells, is consistent with the pH dependence observed for vesicle lipid mixing by the resonance energy transfer assay (Fig. 6).

Overall, these data indicate that the DOPE–N-succinyl-DOPE and DOPE–N-glutaryl-DOPE formulations destabilize under endosomal conditions and simultaneously destabilize the associated endosomal membrane to achieve the release of encapsulated antibiotic. Given the properties of charged vesicles containing high concentrations of unsaturated PE (8), it is likely that neutralization of the anionic lipid at low pH destabilizes the vesicle structure (22, 35). As the negative charge is reduced the tendency for DOPE to adopt nonbilayer phases increases. Such structures have been implicated in a variety of fusion-of-membrane perturbation events including endosome disruption (21). Whether this membrane destabilization is a true membrane fusion event or is a less well characterized process is not known. However, this interpretation is consistent with the data presented in Fig. 6C, which corroborate the observations of Nayar et al. (35) that calcium (20 mM) and/or pH 4.0 can induce the formation of the nonbilayer hexagonal-II phase in DOPE–N-succinyl-DOPE (7/3) mixtures. Work reported by Nayar and Schroit (34) showed that DOPE–N-succinyl-DOPE vesicles were more leaky under acidic conditions than at neutral pH (determined by a dye leakage assay), which they hypothesized may be due to lipid packing defects that arise when the N-succinyl-DOPE is protonated. However, leakage of gentamicin from DOPE–N-succinyl-DOPE liposomes was not observed in the presence of either 10% serum or DMEM (which contains 2 mM Ca²⁺ and in which cells were exposed to liposomal drug) (Table 1). Consequently, it is unlikely that drug leakage occurred from the liposomes in the DMEM solution prior to the endocytosis by the macrophages; otherwise, there would be no improvement in the intracellular activity above that of free drug. Rather, it is likely that the liposomes containing encapsulated gentamicin and in fluid-phase solution (containing 2 mM calcium) were endocytosed and were subsequently destabilized intracellularly by the reduction of the endosomal pH.

In summary, the results presented here demonstrate that gentamicin encapsulation in lipid vesicles that undergo pH-dependent lipid mixing and fusion confers to this membrane-impermeative antibiotic a significant increase in therapeutic activity against intracellular bacterial infections. The pH-sensitive, anionic vesicles described here are presumed to perturb the endosomal membrane following endocytosis and acidification and are expected to be useful for the cytoplasmic delivery of other membrane-impermeative drugs with intracellular sites of action. Experiments designed to evaluate the in vivo therapeutic activities of these carriers are ongoing.