Phospholipid bilayers in model membranes and drug delivery systems: from physics to pharmacy

Antimicrobial peptides (AMPs) emerged as an interesting alternative to antibiotics that fight against infectious diseases. AMPs are considered membrane-active agents leading to cell death by acting on the phospholipid membrane. All proposed bactericidal mechanisms have the same main initialisation – adsorption of AMPs onto the membrane due to electrostatic interactions between the cationic peptides and the accessible anionic groups of hydrophilic phospholipid headgroups at the membrane surface. Thereafter, accumulation and positional change eventually lead to the formation of pores, membrane permeabilisation or its micellisation (Teixeira et al., 2012). The interaction must be selective regarding the distinction between mammalian cells and pathogen cells (bacteria, fungi, protozoa). Thus, knowledge of the details of AMPs interaction with lipid bilayers of composition mimicking the pathogen membrane is of great importance to evaluate their antimicrobial activity. Small-angle X-ray and neutron scattering (SAXS and SANS, respectively) studies can give the necessary information on the AMPs’ ability to affect the structure and integrity of the membrane.

Figure 1

Figure 2

Figure 3

We studied the interaction of cecropin A–melittin (CAM, nominal charge +6), a hybrid peptide composed of the cationic region of cecropin A and the hydrophobic and non-haemolytic region of melittin with a bacterial model lipid membrane composed of zwitterionic palmitoyl-oleoyl-glycerophosphoethanolamine (POPE) and negatively charged palmitoyl-oleoyl-phosphoglycerol (POPG). SANS and SAXS were used to unravel the mechanism of the peptide antimicrobial activity as described in Silva et al. (2018). Fig. 2A depicts the normalised SANS intensity as a function of the scattering vector q of POPE/POPG oligolamellar vesicles with the lipid bilayer thickness being d_L ~ 39.4 Å. SANS data were analysed using SasView software. Fig. 2B shows the I(q) of CAM–POPE/POPG mixtures at P:L = 1:10 mol/mol and 36 °C. When the vesicle dispersion is mixed with the CAM peptide, a fine white precipitate spontaneously forms, suggesting massive aggregation and condensation of the lipid bilayers induced by the peptide. CAM interacts strongly with the negatively charged bilayer and induces extensive vesicle disruption, forming a condensed ‘onion-like’ multilamellar structure with repeat distance d = 2π/q₀ ~ 50.0 Å as derived from the maximum of the observed peak and confirmed by SAXS (not shown). Fig. 3 shows a sketch of the proposed multilamellar structure of CAM–POPE/POPG mixture. We propose that this peptide exerts its antimicrobial action through extensive membrane disruption, leading to cell death.

In another study (Silva et al., 2013), the bilayer composed of dimyristoylphosphatidylcholine/dimyristoylphosphoglycerol (DMPC/DMPG = 3:1 mol/mol) mimicking the membrane of Candida albicans interacted with AMP of the lactoferrin family (LFchimera, nominal charge +12). Fig. 4 shows the SAXS pattern of two bicontinuous cubic phases detected due to disintegration of the membrane by pore-forming mechanism. These results illustrate variedness in the antimicrobial mechanism of AMPs acting on the lipid membrane.

Figure 4

Lipoplexes are formed due to the electrostatic interaction between positively charged liposomes and DNA polyanion. Positively charged complexes show enhanced interaction with the negatively charged cytoplasmic membrane and, therefore, higher cellular uptake via endocytosis. However, for successful DNA delivery, lipoplexes must escape from the endosome inside the cell and DNA must be released from the complex in cytoplasm. High positive surface charge density of the lipoplex is crucial for a successful endosomal fusion. On the other hand, it can be an obstacle for DNA release from the complex. A way to overcome this problem is to use pH-sensitive surfactants with pK_a between 4.5 and 8. Acidic pH inside of endosome secures lipoplexes’ fusion with the endosomal membrane. In the cytoplasm, at neutral pH, the surfactants are uncharged and DNA can be easily released from the lipoplex via dissociation.

We prepared pH-sensitive liposomes composed of homologues of series of N,N-dimethyl-alkane-1-amine N-oxides (CnNO, n = 8–18, where n is the number of carbon atoms in the alkyl substituent) (Devínsky et al., 1978) and neutral phospholipid dioleoylphosphatidylethanolamine (DOPE) at a molar ratio CnNO/DOPE = 0.4 and tested them for in vitro transfection activity. Several techniques (SAXS, UV–VIS, zeta potential measurements, confocal microscopy) were applied to characterise the system in an effort to unravel the relationship between the transfection efficiency, structure and composition of the lipoplexes (for details, see Liskayová et al., 2019). Fig. 5 summarises the most important findings underlining the connection between structural changes (studied by SAXS) and transfection activity.

In acidic conditions, DNA–C8NO/DOPE at C8NO/DOPE = 0.4 mol/mol forms a condensed inverted hexagonal phase (H_II^C). In this structure, the DNA strands are inserted in hexagonally arranged tubules created by the surfactant/lipid mixture (Fig. 1B). Note that the complexes prepared with C8NO keep their structure in neutral/alkaline solutions also. CnNOs with longer hydrophobic alkyl substituent, n ≥ 10, induce structural changes. SAXS revealed the coexistence of two phases, condensed lamellar L_α^C phase and H_II^C in the complexes DNA–CnNO/DOPE, n ≥ 10. Commensurate lattice parameters, the repeat distance d_LC of L_α^C phase and the lattice parameter of the H_II^C phase (d_LC ≈ a_HC) indicate an epitaxial relationship, we abbreviate the structure as L_α^C & H_II^C. The connection through the common scattering plane facilitates the transition of the lipid between the two different structures. The structure changed in neutral/slightly alkaline solutions. A lamellar phase (L_α) is a dominant structure in all studied mixtures. In addition, SAXS patterns of complexes DNA-CnNO/DOPE, n=16 and 18, indicate the presence of a bicontinuous Pn3m cubic phase. The lattice parameter a_Q was found to be a_Q ~ 17–18 nm for detected cubic phases. Bicontinuous cubic phases (Q_II) are formed by a pair of interpenetrating, but non-contacting aqueous channels separated by a single, continuous lipid bilayer. Even if the lattice parameter of Pn3m is rather big (17–18 nm), three-dimensional arrangement of lipid bilayer in the cubic network requires its high negative curvature that might bring difficulties to accommodate bulky DNA molecules. To summarise, the change of pH from acidic to neutral induces phase transition L_α^C&H_II^C → Q_II+L_α, which allows DNA release from the lipoplexes.

Figure 5

Transfection efficiency for plasmid DNA (pDNA, EGFP-N1) was tested on human bone osteosarcoma epithelial cells (U2OS line) and evaluated after 24 and 48 hours by flow cytometry. Commercially available Lipofectamine 2000 (LF) was used as a control. Transfection efficiency follows a quasi-parabolic dependence on the length of CnNO alkyl substituent n, with the maximum found at n = 16. The complex of C16NO/DOPE was found to be ~3 times more efficient than LF. Our findings support the hypothesis that the lipids of Q_II phase favour membrane pore formation resulting from fusion of the lipoplex cubic structure with endosomal membrane. Pores, in turn, allow for cytoplasmic gene delivery. Due to pH sensitivity of our lipoplexes, we can hypothesise synergy of two processes enhancing transfection: a massive release of DNA during L_α^C&H_II^C → Q_II + L_α phase transition and its easier internalisation in the cell supported by the propensity of Q_II for pore formation.

Pulmonary surfactant (PS) is a surface active film lining the alveoli of the lung (Fig. 6A). Its principal function is to lower the surface tension at the air/liquid interface, facilitate the exchange of gasses and stabilise alveoli during breathing. PS is composed of ~90% lipids and 8%–10% of a few specific surfactant-associated proteins. Phospholipids predominate; saturated dipalmitoylphosphatidylcholine (DPPC) is the most abundant in PS of mammals, up to ~50% by mass. Unsaturated phosphatidylcholines (PC) create ~20% and a smaller fraction (~10%) of anionic species such as POPG. A high content of DPPC plays a key function to reach the necessary, extremely low surface tension (<1 mN/m). Hydrophilic SP-A, SP-D and hydrophobic SP-B, SP-C are surfactant-associated specific proteins. SP-B is crucial for proper spreading of the lipid interfacial monolayer. The surface active function of PS is critically dependent on the existence of a multilayered film (~100 nm in thickness) at the air–water interface. An inactivation, deficiency or an absence of PS results in respiratory distress syndrome (RDS) that can be lethal for premature babies. RDS is therapeutically treated by application of PS preparations obtained from animals (exogenous natural), like porcine Curosurf^® or bovine Survanta^®.

The exogenous natural PS contains at least 50 different phospholipids and a small fraction of hydrophobic proteins (~1–2 wt%) (Calkovska et al., 2016). Morphologically, it is a mixture of various vesicles, from unilamellar through oligolamellar up to multilamellar.

We studied the effect of bacterial toxin, lipopolysaccharide (LPS), on the structure of clinically modified porcine pulmonary surfactant, mimicking pathological conditions. Consecutively, Polymyxin B (PxB), a cyclic amphiphatic antibiotic, was applied. Fig. 6B shows the SAXS patterns of modified porcine surfactant at 37 °C when the mixture is in liquid crystalline state (fluid phase). Two peaks (L1 and L2) belong to a lamellar phase with repeat distance d ~ 8.35 nm. LPS affects the multilamellar packing of PS. Longer incubation (~2 hours) generates a large swelling up, which corresponds to a SAXS pattern of large unresolved peak of low intensity with a subtle shoulder at low q ~ 0.5 nm⁻¹. The pattern indicates swelling of a lamellar phase up to periodicity ~12–13 nm. We attribute these structural changes to the PS surface charge unbalance due to LPS insertion. Consecutively, damaged surfactant/LPS 10% was incubated with peptide-based antibiotic PxB. We found that even a small dose of the cationic antibiotic reduces the repeat distance. PS/LPS incubated with 0.5% of PxB for 30 min reduced the repeat distance to d ~ 8.6 nm (not shown). Cationic molecule of PxB acts as an inhibitor of structural disarrangement induced by LPS and restores original lamellar packing. Structural changes were confirmed also by optical microscopy.

Figure 6

The study was focused on the mechanism underlying the structural changes in surface-reducing features (for details, see Kolomaznik et al., 2018). The function of ‘infected’ and consecutively ‘treated’ PS was tested with pulsating bubble surfactometer. The infected PS (surfactant/LPS 10%) was not able to reach the necessary physiologically relevant low surface tension. Intriguingly, the minimal surface tension decreased alongside with structural recovery when applying PxB (Kolomaznik et al., 2018). The obtained results accurately reflect the situation with a native lung surfactant as confirmed by a recent in vivo study (Calkovska et al., 2021) and support the idea of PxB/Curosurf combined therapy in neonatal medicine.

Amphiphilic molecules of lipids and surfactants self-assemble into structures of various morphologies ranging from micelles, lamellar and non-lamellar phases to microemulsions. Each of these morphologies possesses its own unique physical properties and offers the possibility for pharmaceutical applications. In the text above, we demonstrated just a few examples of ‘usability’ of phospholipid bilayers: as a model membrane mimicking the lipid bilayer of bacterial membrane in an effort to unravel the mechanisms of AMPs’ antimicrobial activity, as a carrier designed for genetic material delivery, and finally, a lipid mixture extracted from animal's lungs that is clinically used in neonatal medicine. Findings presented here result from the collaboration between several institutions and have been published to the full extent in Kolomaznik et al. (2018), Liskayová et al. (2019), Silva et al. (2013) and Silva et al. (2018).