Chemical structures of amphotericin B (AmB), carboxymethylcellulose (CMC), poly(diallyldimethylammonium chloride) (PDDA) and the cationic lipid dioctadecyldimethylammonium bromide (DODAB) are on Table 1. DODAB self-assembly in water dispersion yields bilayer fragments (BF) by ultrasonic input with a macrotip probe.

The existence of bilayer fragments from synthetic lipids such as sodium dihexadecylphosphate, or dioctadecyldimethylammonium bromide or chloride obtained by sonication with tip has been supported by the following evidences: (i) osmotic non-responsiveness of the dispersion indicative of absence of inner vesicle compartment 29; (ii) TEM micrographs with electronic staining 30; (iii) cryo-TEM micrographs 31; (iv) fluid and solid state coexistence and complex formation with oppositely charged surfactant 32; (v) solubilization of hydrophobic drugs at the borders of DODAB bilayer fragments, which does not occur for DODAB closed bilayer vesicles 1920213334. They differ from the closed vesicles by providing hydrophobic borders at their edges that are absent in closed bilayer systems such as vesicles or liposomes. Under conditions of low ionic strength, due to electrostatic repulsion, the charged bilayer fragments remain colloidally stable in aqueous dispersions 1920213334.

In fact, DODAB BF have been used to solubilize AmB 19 at room temperature as schematically shown in Figure 1. This solubilization takes place at low drug-to-lipid molar proportions (low P) and presents certain limitations: 1) hydrophobic edges of bilayer fragments have a limited capacity of solubilizing the hydrophobic drug; 2) the bilayer core in the rigid gel state is too rigid to allow solubilization of AmB at room temperature being a poor solubilizer for this difficult, hydrophobic drug 19203335. On the other hand, at high P, AmB aggregates in water solution can be considered as drug particles. These can be surrounded by a thin cationic DODAB bilayer as previously described 35 (Figure 1).

The physical properties of different dispersions such as size and zeta-potential are given in Table 1 both at low and high P. The drug in water exhibits substancial aggregation (Dz = 360–433 nm), as expected from its hydrophobic character. The drug particle presents a negative zeta-potential of -26 mV explained by dissociation of its carboxylate moiety at the pH of water 35. Upon changing the medium to IGP buffer, as previously reported, a decrease in size for AmB aggregates was observed (Dz = 75 nm) (Table 1), due to the chaotropic (dispersing) effect of dihydrogenphosphate anion on AmB aggregates 35. Both types of AmB aggregates interacted with DODAB BF yielding either loaded BF fragments at low P or DODAB covered drug particles at high P. The characteristics of these cationic assemblies before and after their interaction with oppositely charged CMC over a range of concentrations (0.001–1.0 mg/mL) are in Table 1. At low P, charge reversal took place above 1 mg/mL CMC whereas at high P, it occurred above 0.1 mg/mL CMC (Table 1).

At low P, the effect of CMC concentration on DODAB BF/CMC (unloaded control) or DODAB BF/AmB/CMC properties is in Figure 2. At low P and 1 mg/mL CMC, DODAB BF/AmB/CMC anionic complexes present 90 nm mean diameter and -50 mV of zeta-potential. The low size and large surface potential mean high colloid stability, so that this was the condition chosen for coverage with cationic polyelectrolytes. In the presence of CMC, there are two regions of colloid stability for cationic or anionic assemblies characterized by small sizes: regions I and III, and one region of instability: region II, characterized by aggregation and large sizes (Figure 1). Charged particles covered by oppositely charged polyelectrolytes exhibited similar profiles for the colloid stability as a function of polyelectrolyte concentration 3637.

The aggregation state of AmB at low P was evaluated from optical spectra (Figure 3). The drug in DMSO:methanol 1:1 yields a spectrum of completely solubilized, nonaggregated drug since this organic solvent mixture is the one of choice for AmB solubilization (Figure 3A). The drug in water exhibits the typical spectrum of aggregated AmB (Figure 3B). As depicted from AmB spectrum in DODAB BF (Figure 3C) or DODAB BF/AmB/CMC (Figure 3D), the drug is found in its monomeric state and completely solubilized. In fact, solubilization of AmB in DODAB BF, at low P, was previously described 19. This formulation employing DODAB BF at low P was very effective in vivo 21 and exhibited low nephrotoxicity 22.

At low P, the effect of [PDDA] on sizes and zeta-potentials of DODAB BF/AmB/CMC assemblies at 1 mM DODAB, 0.005 mM AmB and 1 mg/mL CMC is on Figure 4. The region of PDDA concentrations for size minimization and high colloid stability was very narrow and around 1 mg/mL PDDA. Below and above this concentration, about 300 nm and negative zeta-potentials, or 500–700 nm of zeta-average diameter and positive zeta-potentials were obtained, respectively (Figure 4). Size minimization at Dz = 171 nm and zeta-potential = 24 mV for the DODAB BF/AmB/CMC/PDDA assembly was not related to optimal fungicidal activity as depicted from the 79% of C. albicans viability (Table 2). Possibly, the total positive charge on the assembly was not sufficient to substantially reduce fungus viability. For final coverage with polylysines (PL) of increasing molecular weight at 1 mg/mL PL, there was an increase in the final zeta-potential modulus and a larger loss of viability (Table 2). The DODAB BF/AmB/CMC/PDDA formulation at low P was 100% effective against the fungus only at 5 mg/mL PDDA (Figure 5D).

The importance of large positive zeta-potentials for high efficiency of drug assemblies with DODAB BF and polyelectrolytes can be clearly seen from Figure 5. Negatively charged assemblies like those in Figure 5A and 5B yielded 100% of cell viability. Positively charged assemblies obtained upon increasing [PDDA] reduced cell viability to 50% (CMC/PDDA) (Figure 5C) or to 0% (DODAB BF/AmB/CMC/PDDA above 5 mg/mL PDDA) (Figure 5D). The schematic drawing in Figure 5D illustrates the layered assembly of microbicides in a single supramolecular assembly. The first attack comes from the outer cationic polyelectrolyte layer. Upon unwrapping this first layer and the second inert CMC layer, monomeric AmB contacts the fungus cell followed by the also effective DODAB action. Maybe this design represents a very effective assembly against multidrug resistance. Complete loss of fungus viability can seldom be achieved at the same separate doses of each component 25.

The complexation between DODAB BF and CMC was previously studied in detail by our group 36. DODAB BF at 0.1 mM DODAB and CMC (0.001–2 mg/mL) are, in fact, electrostatically driven to complexation from the electrostatic attraction (Figure 6A and 6B).

At high P, 0.1 mM DODAB BF is sufficient to cover all AmB particles present in dispersion at 0.05 mM AmB with a thin, possibly bilayered, 6–8 nm DODAB cationic shell as previously described 35. This cationic interface is expected to interact with the oppositely charged CMC polyelectrolyte. At 0.1 mg/mL CMC, AmB/DODAB BF/CMC anionic complexes present high colloid stability, 230 nm mean diameter and -34 mV of zeta-potential (Figure 6C and 6D). This condition was chosen for further coverage with cationic polylectrolytes.

Regarding the aggregation state of AmB, as expected, at 0.05 mM AmB, the majority of drug molecules were found in the aggregated state. Spectra in IGP buffer (Figure 7A), after drug particle coverage with 0.1 mM DODAB BF (Figure 7B) or with 0.1 mM DODAB BF plus 0.1 mg/mL CMC (Figure 7C) revealed the typical profile of aggregated drug. The spectrum in Figure 7C indicates a certain amount of monomeric drug not present in the other spectra (Figure 7A and 7B). Possibly, CMC sterically stabilized DODAB BF preserving hydrophobic sites of DODAB BF to be occupied by the monomeric drug. In absence of CMC, DODAB BF might fuse diminishing drug solubilization at their rim.

At the chosen condition for the AmB/DODAB BF/CMC assembly, the effect of increasing [PDDA] was an initial colloid stabilization (decrease in size) around 1 mg/mL PDDA followed by further destabilization (increase in size) above this concentration (Figure 8A), possibly due to bridging flocculation 38. Zeta-potential displayed the usual sigmoidal dependence on [PDDA] (Figure 8B).

The importance of positively charged assemblies at high P for fungicidal activity is emphasized in Figure 9. C. albicans remains 100% viable in the presence of negatively charged CMC only (Figure 9A), 70% viable in the presence of negatively charged AmB/DODAB BF/CMC at high P (Figure 9B), 50–60% viable in the presence of CMC/PDDA at [PDDA] > 1 mg/mL and 0% viable in the presence of AmB/DODAB BF/CMC/PDDA at PDDA ≥ 2 mg/mL (Figure 9D).

Alternatively, PDDA was replaced by PL (Table 2). At high P, the effect of increasing PL molecular weight was an increase in size, an increase in zeta-potential and a decrease of % of cell viability (Table 2). Table 1 summarized the different properties of assemblies at low and high P. One should notice that coverage of a drug particle with a thin DODAB layer led to a positive zeta-potential of only 9 mV. CMC was slightly attracted to the covered particle producing a looser assembly than the one obtained with CMC coverage of DODAB BF, where electrostatic attraction is due to a higher zeta-potential on the bilayer fragments, typically 41 mV. The particles are loosely or tightly packed depending on the electrostatic attraction between oppositely charged components (cationic layer and CMC) depicted from zeta-potentials. This certainly made a large difference for occasion of drug delivery to the fungus cell. Having a loose or a more tightly packed assembly originated considerable differences in the profile of cell viability as a function of zeta-potential (Figure 10). For the less tightly packed drug particles at high P, drug delivery was more efficient leading to drug release and cell death at lower zeta-potentials (Figure 10). The reason for this high efficiency at low zeta-potential is associated both with the high P, meaning high drug dose, and with the loosely packed nanoparticle assembly.

Fungizon (AmB in deoxycholate) and DODAB BF/AmB (formulation at low P) were previously evaluated in mice with systemic candidiasis 21. Both formulations yielded equivalent therapeutic results. However, DODAB BF/AmB was better from the point of view of reduced nephrotoxicity 22. Furthermore, cationic surfactants and polymers have an effect on integrity of red blood cells 28. Therefore, similar studies should be performed for the formulations described in this paper.

Dioctadecyldimethylammonium bromide (DODAB), 99.9% pure was obtained from Sigma Co. (St. Louis, MO, USA). Carboxymethyl cellulose sodium salt (CMC) with a nominal mean degree of substitution (DS) of 0.60–0.95, poly(diallyldimethylammonium chloride) (PDDA) with M_v 100,000–200,000 and polylysines (PL) were obtained from Sigma (Steinheim, Germany) and used without further purification. Amphotericin B (AmB, batch 008000336) was purchased from Bristol-Myers Squibb (Brazil) and was initially prepared as a 1 g/L stock solution in DMSO/methanol 1:1. Candida albicans ATCC 90028 was purchased from American Type Culture Collection (ATCC) and reactivated in Sabouraud liquid broth 4% before plating for incubation at 37°C/24 h. In order to prepare fungal cell suspension for antifungal activity assays, three to four colonies were picked from the plate and washed twice either in isotonic glucose phosphate buffer (IGP; 1 mM potassium phosphate buffer, pH 7.0, supplemented with 287 mM glucose as an osmoprotectant) 3940 or in Milli-Q water by centrifugation (3000 rpm/10 minutes), pelleting and resuspension. The final fungal cell suspension was prepared by adjusting the inoculum to 2 × 10⁷ cfu/mL and then diluting by a factor of 1:10 either in IGP or in Milli-Q water yielding 2 × 10⁶ cfu/mL.

DODAB was dispersed in water or IGP buffer, using a titanium macrotip probe 41. The macrotip probe was powered by ultrasound at a nominal output of 90 W (10 minutes, 70°C) to disperse 32 mg of DODAB powder in 25 mL water solution. The dispersion was centrifuged (60 minutes, 10000 g, 4°C) in order to eliminate residual titanium ejected from the macrotip. This procedure dispersed the amphiphile powder in aqueous solution using a high-energy input, which not only produced bilayer vesicles but also disrupted these vesicles, thereby generating open BF 2941. Analytical concentration of DODAB was determined by halide microtitration 42 and adjusted to 2 mM.

Stock solutions of AmB were prepared at 1 mg/mL in 1:1 DMSO/methanol. Stock solutions of PDDA, CMC and PL were prepared at 20 mg/mL and diluted in the final dispersion to yield the desired final concentration. The stock solution of AmB (1 mg/mL) was added to DODAB BF dispersions to yield low and high drug to lipid molar proportions (P). At low P, dispersions contained final concentrations of drug, DODAB, CMC and PDDA equal to 0.005 mM (5 micrograms/mL), 1 mM (631 micrograms/mL), 0.01–2.00 mg/mL and 0.01–10.00 mg/mL, respectively. Firstly, DODAB BF and drug were allowed to interact for 10 minutes. Thereafter, CMC was added and allowed to interact for 20 minutes before adding PDDA, which was also allowed to interact for 20 minutes, before determining zeta-average diameter and zeta-potentials. At high P, a similar procedure was done this time at final concentrations of drug, DODAB, CMC and PDDA equal to 0.050 mM (50 micrograms/mL), 0.1 mM (63.1 micrograms/mL), 0.01–2.00 mg/mL and 0.01–10.00 mg/mL, respectively. At high P, drug particles were obtained at 0.050 mM AmB in IGP buffer yielding particles with 75 nm zeta-average diameter and -27 mV zeta-potential 35. These drug particles were firstly covered by DODAB BF and then wrapped by the polyelectrolytes over the quoted range of concentrations. Sizes and zeta-potentials were determined by means of a ZetaPlus Zeta-Potential Analyser (Brookhaven Instruments Corporation, Holtsville, NY, USA) equipped with a 570 nm laser and dynamic light scattering at 90° for particle sizing 43. The zeta-average diameters referred to in this work from now on should be understood as the mean hydrodynamic diameters D_z. Zeta-potentials (ζ) were determined from the electrophoretic mobility μ and Smoluchowski's equation, ζ = μη/ε, where η and ε are medium viscosity and dielectric constant, respectively. All D_z and ζ were obtained at 25°C, 1 h after mixing.

UV-visible optical spectra (280–450 nm) for characterization of AmB aggregation state were obtained in the double-beam mode by means of a Hitachi U-2000 Spectrophotometer against a blank of DODAB BF or DODAB BF/CMC (without drug), to separate light scattered by the dispersions from light absorption by the drug. All spectra were obtained at room temperature (25°C) at about 20 minutes after mixing DODAB BF and AmB at low or high drug to lipid P or after adding CMC to DODAB BF/drug assemblies.

At low or high P, DODAB/drug assemblies were wrapped by two layers of oppositely charged polyelectrolytes so that cfu were counted as a function of CMC and/or PDDA concentrations at 1 h of interaction time between C. albicans (1 × 10⁶ cfu/mL) and formulations. Plating on agar plates for cfu counts was performed by taking 0.1 mL of a 1000-fold dilution in Milli-Q water of the mixtures. After spreading, plates were incubated for 2 days at 37°C. CFU counts were made using a colony counter. At low P, final DMSO/methanol concentration is 0.5% whereas at high P it is 5%. No effect of the solvent mixture at 0.5% on cells viability was previously detected 25. For further studies in vivo and at high P, it will be advisable to perform a dialysis step for the cationic nanoparticles aiming at complete elimination of the toxic solvent mixture.