Assessment of free fatty acids and cholesteryl esters delivered in liposomes as novel class of antibiotic

Healthcare associated infections (HAI) with multidrug-resistant (MDR) bacteria continue to be a global threat, highlighting an urgent need for novel antibiotics. In this study, we assessed the potential of free fatty acids and cholesteryl esters that form part of the innate host defense as novel antibacterial agents for use against MDR bacteria. Liposomes of six different phospholipid mixtures were employed as carrier for six different fatty acids and four different cholesteryl esters. Using a modified MIC assay based on DNA quantification with the fluoroprobe Syto9, formulations were tested against Gram-positive and Gram-negative bacteria implicated in HAI. Formulations with MIC values in the low μg/mL range were further subjected to determination of minimal bactericidal activity, hemolysis assay with sheep erythrocytes, and cytotoxicity testing with the human liver cell line HepG2. The potential for synergistic activity with a standard antibiotic was also probed. Palmitic acid and stearic acid prepared in carrier 4 (PA4 and SA4, respectively) were identified as most active lipids (MIC against MDR Staphylococcus epidermidis was 0.5 and 0.25 μg/mL, respectively; MIC against vancomycin resistant Enterococcus faecalis (VRE) was 2 and 0.5 μg/mL, respectively). Cholesteryl linoleate formulated with carrier 3 (CL3) exhibited activity against the S. epidermidis strain (MIC 1 μg/mL) and a Pseudomonas aeruginosa strain (MIC 8 μg/mL) and lowered the vancomycin MIC for VRE from 32–64 μg/mL to as low as 4 μg/mL. At 90 μg/mL PA4, SA4, and CL3 effected less than 5 % hemolysis over 3 h and PA4 and CL3 did not exhibit significant cytotoxic activity against HepG2 cells when applied at 100 μg/mL over 48 h. Our results showed that selected fatty acids and cholesteryl esters packaged with phospholipids exhibit antibacterial activity against Gram-positive and Gram-negative bacteria and may augment the activity of antibiotics. Bactericidal activity could be unlinked from hemolytic and cytotoxic activity and the type of phospholipid carrier greatly influenced the activity. Thus, fatty acids and cholesteryl esters packaged in liposomes may have potential as novel lipophilic antimicrobial agents.


Staphylococcus aureus (MRSA), vancomycin-resistant
Enterococcus faecalis (VRE), Pseudomonas aeruginosa, and Acinetobacter baumannii [2]. Multidrug-resistant Staphylococcus epidermidis has become notorious for catheter associated infections. Nationwide, at least one case of HAI occurs for every 25 patients [3]. Despite some progress in containing the occurrence of HAI through infection control measures [4], an urgent need for the development of novel antibacterial agents remains. No new class of antibiotic has been discovered since 1987 and the last six approved new drugs are all modifications of existing drugs [5,6]. Drug discovery through target-focused screening of large libraries of synthetic compounds has failed with a major reason being insufficient compound penetration into the microbial cell [7], and transition of natural antimicrobial peptides to the clinical market has proven to be difficult [8,9]. Ongoing efforts include antisense RNA targeting essential microbial genes [10] and exploitations of uncultured bacteria [11]. However, host derived lipids, increasingly recognized as important antimicrobial effectors of the innate host defense [12,13], might also be exploited for novel drug design.
Lipids are a widely heterogeneous group of molecules that share hydrophobic or mixed hydrophobic/hydrophilic properties [14]. The major lipid classes include: fatty acids-the basic building blocks for more complex lipids, cholesteryl esters, and phospholipids. Cholesteryl esters are formed through esterification of a fatty acid to cholesterol. Phospholipids typically consist of a glycerol with two fatty acid residues attached, a phosphate group and additions such as choline, an alcohol, or amines. The antimicrobial activity of fatty acids has been known for some time (reviewed in [15]) while antimicrobial properties of cholesteryl esters have been more recently discovered [16].
Motivated by the urgent need for novel antibacterial strategies, the main objective for this study was to examine whether lipids identified as antibacterial effectors in body fluids may have a potential to be used in innovative drug design. Due to their inherent hydrophobicity, lipids require carriers in aqueous environments. Liposomes have been used in the past for delivery of compounds with hydrophobic properties [17] offering in some cases the benefit of reducing cytotoxicity [18]. Liposomes are phospholipid vesicles where the polar head groups of phospholipids interact with the aqueous outer environment and the fatty acid tails provide a hydrophobic environment for lipophilic compounds. When used as carrier, liposomes also provide the advantage of allowing incorporation of tags for targeted delivery [19,20].

Antibiotics
Oxacillin, cefotaxime, vancomycin, tetracycline, and ciprofloxacin were purchased from Sigma-Aldrich in salt form, dissolved in dH 2 O, and kept as stock solution (10 mg/mL) at −20 °C.

Liposome preparation
Six different carrier liposome formulations (carrier-1-6) were generated using the phospholipids and cholesterol  [21]. Liposomes were prepared by dissolving the carrier and test lipids (0.8:0.2, respectively) in chloroform/ methanol (1:1 v/v). The organic solvents were evaporated under a stream of nitrogen gas at 65 °C, and residual solvent removed in vacuo for >24 h. Unilamellar liposomes were formed by hydrating the lipid films with 9 % sucrose and probe sonicating until translucent. The liposomes were then sterilized by passage through a 0.22 µm filter. Each liposome preparation was characterized by dynamic light scattering (UPA 150, Microtrac, Montgomeryville, PA; see Additional file 1: Table S1). Liposomes were stored in 9 % sucrose at 1 mg/mL test lipid (5-10 mg total lipid/mL) for up to 6 month at 4 °C. As a control 9 % sucrose was used. When necessary, liposomal formulations containing test lipids were diluted with 9 % sucrose to ten-fold concentrated working stocks. Stock solutions for the respective carrier liposomes without test lipids were diluted so that the phospholipid concentration was the same as the phospholipid concentration in the formulations with test lipid.

Determination of minimal inhibitory concentration (MIC)
Following Clinical and Laboratory Standards Institute M07-A9 for broth microdilution assay [22], 10 μL aliquots of serially diluted 10-fold concentrated antibiotic were added to general assay 96 well-round bottom polystyrene microtiter plates followed by the addition of 90 μL of diluted bacteria as prepared in above. For sterility control, 1.1-fold cation-adjusted Mueller-Hinton broth (MHB+) without bacteria was included in the testing. The plates were then incubated at 37 °C for 16-24 h, as recommended, in air and turbidity was read at 650 nm. MIC was the lowest concentration of antibiotic that prevented growth reflected in measurable turbidity.

Modified minimal inhibitory concentration assay
Due to the inherent turbidity of the liposomal formulations interfering with absorbance readings, the MIC protocol was modified; after completion of the MIC assay the DNA content was quantified with the fluoroprobe Syto9 (Invitrogen, Carlsbad, CA). Syto9 freely crosses cell membranes and binds to DNA upon which green fluorescence is emitted. The degree of fluorescence correlates with the amount of DNA, thus with the bacterial concentration in a sample, regardless of sample turbidity contributed by the liposomal formulations [16]. Using black untreated 96 well microtiter plates with clear flat bottom wells (Costar, Corning Life Sciences, Union City, CA) bacteria were otherwise prepared, mixed with antibiotics or liposomes, and incubated according to the standard MIC procedure. Each assay included a growth control (incubation of bacteria in MHB+) and a killing control (incubation of bacteria with effective antibiotic). In addition, liposomes and antibiotics were incubated in the absence of bacteria to control for sterility of the reagents and to obtain background fluorescence readings. At the end of the incubation period, 100 μL of 1.25 μM Syto9 (diluted in dH 2 O) was added to each well. After 15 min incubation at RT in the dark, relative fluorescence units (RFU) were measured (485 nm ex /530 nm em ; Tecan GENios, Tecan Systems, Inc., San Jose, CA). RFUs obtained from liposomal formulations or antibiotic incubated in medium without bacteria (background fluorescence) were subtracted from the RFUs obtained from corresponding wells with bacteria present. All liposomal formulations were first screened at 64 μg/mL test lipid. Those formulations that, compared to untreated bacteria, effected at least 95 % reduction of RFUs of at least two different bacterial strains were further investigated. They were subjected to serial dilution and determination of modified MIC. Modified MIC was defined as the lowest concentration of liposomal formulation that produced RFUs at background level.

Determination of minimal bactericidal concentration (MBC)
After completion of the MIC assay, 6 µL of each well was serially diluted and spot plated. Based on Clinical and Laboratory Standards Institute guidelines M26-A [23], concentrations of liposomal formulations that reduced growth to less than 15 colony forming units (CFU)/6 μL or 2500 CFU/mL were considered bactericidal.

Vancomycin synergism
The protocol for the modified MIC assay was employed with the following changes: instead of 10 μL of tenfold concentrated antibiotic or test lipid, 5 µL of 20-fold concentrated antibiotic and 5 µL of 20-fold concentrated test lipid was used. First, the MIC of vancomycin applied alone was established for VRE and then, serially diluted vancomycin was tested in the presence and absence of 5 μg/mL cholesteryl linoleate in carrier 3.

Hemolytic activity testing
Sheep red blood cells (RBCs) at 50 % v/v whole blood in Alsever's solution (Sigma-Aldrich) were washed by repeated centrifugation at 500×g and resuspension in saline (0.9 % w/v) until the absorbance of supernatant at 405 nm equaled the absorbance of saline. On the final wash, RBCs were adjusted to 15 % packed cell volume in normal saline (15 parts sedimented RBC plus 85 parts saline). In microplates, 10 μL of this was then added to 90 μL liposomal formulations or carrier which had been diluted in saline to the desired test concentration. As negative control, RBCs were mixed with saline. To obtain maximal (100 %) hemolysis, Triton X-100 (Fisher Scientific) was added to 1 % final concentration. Samples were then incubated at 37 °C, 7.5 % CO 2 for up to 24 h. At the end of each incubation period (30 min, 3 h, and 24 h), the percent hemolysis was estimated by the quantification of liberated hemoglobin by absorbance reading. To prevent errors due to liposomal components affecting the oxidation state of the hemoglobin (and thus the absorbance reading) and to dissociate all major hemoglobins from other sample components (also affecting the absorbance reading), all hemoglobin species were converted to hematin based on the Alkaline Hematin Detergent-575 method [24]. Briefly, intact cells were pelleted by centrifugation at 4500×g and the resulting supernatants were diluted to 50 % v/v in 1 % Triton-X-100/0.5 M NaOH.

Cytotoxicity testing
Cytotoxic activity of liposomal formulations was assessed with a hepatic cell line that is commonly used for this purpose [25]. . Stocks for carrier liposomes were diluted so that their phospholipid concentration was the same as the phospholipid concentration of the test lipid in its respective formulation. As negative control, 9 % sucrose diluted 1:10 in culture medium was used. After 48 h incubation at 37 °C with 5 % CO 2 cell viability was determined with the XTT assay (Roche Diagnostics Pleasanton, CA) according to the manufacturer's instructions. Previously untreated cells that were incubated for 15 min with 0.1 % Triton X 100 immediately prior to the XTT assay were used as positive control for cytotoxicity. Cytotoxic activity of lipids was quantified based on the inhibition of cell proliferation after lipid treatment relative to untreated cells.

Data analysis
MIC and MBC assays were conducted in duplicates and repeated at least once. Hemolysis assays were performed in triplicate and cytotoxicity assays were conducted in replicates of four and repeated for a total of three independent experiments. Microsoft Excel 2013 was used for raw data analysis and graphing. IBM SPSS Statistics version 20 was used for statistical analysis.

Validation of modified MIC assay with fluorescence readout
Due to their hydrophobicity-in particular cholesteryl esters cannot be dissolved in aqueous media-test lipids were incorporated into carrier liposomes. Six different formulations designated as carrier 1-6 were tested. The particulate nature of liposomes interferes with turbidity readings which are routinely used in standard MIC assays to evaluate bacterial growth in the presence and absence of antibiotics. Therefore, in this study, we employed a modified antimicrobial susceptibility testing assay that utilizes the green fluorescent DNA probe Syto9 to quantify bacterial proliferation [16]. As summarized in Table 2, the modified assay with fluorescence readout produces MICs that were mostly comparable to those determined with standard MIC testing, with both sets of MIC values within acceptable ranges as defined by the Clinical and Laboratory Standards Institute.

Spectrum of activity of liposomal formulations
First, all formulations were screened at 64 μg/mL test lipid (free fatty acids or cholesteryl esters) against bacterial strains that cause HAI and are resistant to multiple antibiotics (Table 3). For comparison, a P. aeruginosa strain previously reported to be susceptible to cholesteryl esters (PACF, [16]) was also used. When test lipids were serially diluted, the corresponding carrier lipids were similarly diluted maintaining the ratio of test lipid to carrier lipid. For each test lipid the corresponding carrier formulation itself was also assayed at the equivalent concentration. Of note, the liposome size was influenced by the lipid species (see Additional file 1: Table S1).
Based on an at least 80 % inhibition threshold, overall, test liposomes were more effective against the Gram-positive strains tested than the Gram-negative strains. Among the Gram-positive bacteria tested, S. epidermidis and VRE were inhibited by several formulations representing both fatty acids, in particular PA4 and SA4, and cholesteryl esters (CL3 and CA4 for S. epidermidis and VRE, respectively). Among the Gram-negative bacteria tested, one of the two P. aeruginosa strains and the A. baumannii strain were completely resistant to all lipids tested. Furthermore, the E. cloacae strain was inhibited only by two fatty acid formulations (LA4, AA4) and one cholesteryl ester formulation (CL3). However, the cystic fibrosis isolate of P. aeruginosa exhibited broad susceptibility to several fatty acid and cholesteryl ester formulations suggesting that there is a strain dependent susceptibility. In general, test lipids prepared with carrier 3 and carrier 4 demonstrated the greatest activity. There was also some inhibitory activity of the carrier lipids independent from the presence of fatty acids or cholesteryl esters (see Table 3).
Those test lipid formulations that effected at least 95 % growth inhibition against at least two different bacterial strains in the screening assay were further subjected to determination of minimal inhibitory concentration ( Table 4). The lowest MIC values were determined for  against PACF in the screening assay was not confirmed by the MIC assay.

Bactericidal activity of selected liposomal formulations
The liposomal formulations PA4, SA4 and CL3 were then subjected to testing for minimal bactericidal concentration (MBC) against S. epidermidis (Fig. 1). PA4 and SA4 demonstrated strong bactericidal activity (MBC of 0.5 and 0.25 μg/mL, respectively). In contrast, CL3 was not bactericidal at the tested concentrations, reducing the bacterial inoculum by no more than 75 %.

Effect of liposomes on vancomycin susceptibility
To address whether antimicrobial lipids in liposomal formulation may have the potential to augment the effects of standard antibiotics, MIC of vancomycin was determined for VRE in the presence and absence of CL3, which alone-or its carrier liposomes-did not effect a substantial growth inhibition (see Table 3), or carrier liposomes. The MIC of vancomycin against VRE shifted from 64-32 μg/mL to as low as 4 μg/mL in the presence of 5 μg/mL CL3 (Fig. 2), which would be consistent with a shift from vancomycin resistant to susceptible, according to CLSI-M100-S22. This suggests that liposomal lipids may potentiate or restore the efficacy of antibiotics against multidrug-resistant bacteria.

Hemolytic and cytotoxic activity of selected liposomal formulations
Prior to considering clinical use, a candidate compound must be evaluated for hemolytic and cytotoxic activity. PA4, SA4, and CL3 were subjected to hemolysis testing (Table 5) and cytotoxicity testing (Fig. 3). At high dose (900 μg/ mL test lipid) PA4 and SA4 effected over 40 % hemolysis within 30 min and CL3 within 3 h. However, when tested at 90 μg/mL test lipid (between 90 and 360 times the lowest observed MIC) none of the formulations exhibited significant hemolysis over 3 h. Carrier-3 and carrier-4, in the absence of test lipid, also exerted hemolytic effects at the higher dose (both over 40 % hemolysis after 3 h), but did not cause hemolysis when tested at 90 μg/mL over 3 h (2.24 ± 0.6 and 1.64 ± 0.2 % hemolysis, respectively).

Table 4 Minimal inhibitory concentrations (MIC, given in μg/mL) of test lipids that effected a growth inhibition of >95 % against at least two different bacterial strains in the screening assay
Shown are MIC values derived from two independent experiments each performed in triplicate. NT not tested because not susceptible in the screening assay a See Table 1  Cytotoxic activity was evaluated with HepG2 cells at 100 μg/mL employing the metabolic XTT assay (Fig. 3). While SA4 had significant inhibitory effects, cells treated with PA4 and CL3 were statistically not different from untreated cells. Carrier-3 and carrier-4, when tested alone, also exhibited cytotoxic effects (40.68 ± 12.54 % inhibition, p < 0.05, and 83.97 ± 13.88 % inhibition, p < 0.001, respectively). This data suggests that the hemolytic and cytotoxic activity of the liposomal formulations is at least in part contributed by the carrier lipid. Considering the varying activities of different lipids incorporated in the same carriers, a complex interaction between the different lipid species is likely.

Discussion
In this study, striving to address the urgent public health need posed by HAI with multidrug-resistant bacteria, we assessed the potential of free fatty acids and cholesteryl esters as novel antibacterial agents. Liposomes of varying composition were employed as carrier for these lipophilic agents and these formulations were tested against Gram-positive and Gram-negative multidrugresistant bacteria involved in HAI using a modified MIC. The potential for restoring activity of conventional antibiotics was probed and hemolytic and cytotoxic activity assayed. Palmitic and stearic acid as well as cholesteryl linoleate were identified as the most promising compounds whereby the carrier liposome formulation influenced their activity.
To overcome interference of turbidity contributed by liposomes we employed the DNA probe Syto9 to quantify bacterial growth in an otherwise standard MIC assay. Styo9 has been often used to quantify viable bacteria in biofilms [26,27] but not yet been validated for use in MIC assays, which we did here using quality control strains and standard antibiotics. The modified DNA probe-based MIC assay may also be useful for testing fastidious organisms that require the addition of blood products in their growth media making turbidity based measurements including standard MIC testing obsolete.  Among the fatty acids and cholesteryl esters tested, palmitic acid, stearic acid, and cholesteryl linoleate emerged as the most active lipids. Palmitic acid and stearic acid acted only against multidrug-resistant S. epidermidis and vancomycin resistant E. faecalis, both representing Grampositive cocci. This is in line with an earlier report on the preferential bactericidal activity of fatty acids against Grampositive bacteria [28] though the stated effective concentrations of palmitic acid were substantially higher than what we describe here. The relatively decreased activity reported by Huang et al. [28] may be attributed to the use of ethanol-solubilized fatty acid without a carrier underlining the potential of liposomal formulations to reduce required active concentrations. Cholesteryl linoleate showed activity against the multidrug-resistant S. epidermidis and a P. aeruginosa strain but was less effective in killing than stearic acid and palmitic acid. This suggests that the cholesteryl ester may have a different mode of action.
We have identified host derived lipids that may have promise for future use as antimicrobials when applied alone or in combination with standard antibiotics. Plant lipids, namely essential oils, have long been used for their antibacterial activity in the food industry [29,30]. Recently, though, they have been also assessed for clinical use [31,32], including the use as enhancer of antibiotics [33]. In a rabbit sepsis model, co-administration of the polyunsaturated fatty acids linolenic and arachidonic acid with antibiotics enhanced killing of a multidrug-resistant P. aeruginosa strain [34]. Adding lipophilic properties to antimicrobial peptides by N-acylation has also been proven to enhance their antimicrobial activity [35,36].
We observed that the carrier liposome composition influenced the test lipid activity and that some formulation exhibited independent antibacterial activity. This suggests that phospholipid-mediated antimicrobial activity could contribute to the low MIC of palmitic acid and stearic acid. An interaction between the carrier lipids and the test lipids is also implicated by the changes in liposomes size depending on the constituents. Possible mechanisms of the observed liposome activity may include fusion and subsequent disruption of the bacterial cell membrane through lipid embedding into the phospholipid bilayer and, for unsaturated lipids, lipid peroxidation with radical production [15]. Such mechanistic studies are currently under way in our laboratory.
Bacterial and mammalian cell membranes both are composed of phospholipid bilayers. Thus, it is conceivable that the antibacterial liposome formulations exhibit hemolytic and cytotoxic activity against mammalian cells. At concentrations ninety times or higher than the lowest MIC no hemolytic activity was observed within the first 3 h but prolonged incubation resulted in hemolysis. This suggests that topical use may be a preferable treatment strategy for liposomal lipids. Cytotoxic effects through lipid peroxidation and mitochondria-initiated apoptosis have been well documented, in particular for unsaturated fatty acids such as linoleic acid [37,38], and cytotoxic fatty acids are under investigation for intended use in cancer treatment [39,40]. Here, we found that palmitic acid and cholesteryl linoleate did not cause significant growth inhibition of HepG2 cells but that stearic acid displayed significant cytotoxicity. Considering that the degree of antibacterial activity was influenced by the carrier composition and that the carriers also exerted some hemolytic and cytotoxic effects, careful selection and modification of the carrier lipids may yield antimicrobial formulations devoid of hemolytic and cytotoxic activity.
Liposomes have been previously used as carrier for antimicrobial compounds. For example, the antifungal drug amphotericin B was successfully formulated in liposomes with greatly reduced cytotoxicity [41]. Other applications include liposomal delivery of antibiotics to pathogens entrapped in biofilms and intracellular pathogens [42,43]. A major advantage of using liposomes as carrier for antimicrobial lipids (and other compounds) is the possibility to insert molecules that enable targeted delivery of the drug [44].