How Does Melittin Permeabilize Membranes?

Main Text

Fifty years ago, the sequence of melittin, the archetypal membrane-permeabilizing peptide, was first determined (1). Soon after, the basic biophysical principles of its activity in membranes were understood. Melittin, which is the major component of Honey Bee venom, partitions strongly into membranes and, by virtue of its structural amphipathicity, disrupts the normally strict segregation between the polar and nonpolar parts of the lipid bilayer and its surroundings, enabling passage of ions, water, small molecules, and, sometimes, macromolecules. Yet, 50 years later, debate continues about the molecular mechanism of action of melittin and other membrane-permeabilizing peptides.

Synthetic lipid bilayers have long been used to study the mechanism of melittin. These studies show that there is not a unique mechanism, but instead a mechanism that varies with experimental conditions. Melittin was initially found to form voltage-dependent, membrane-spanning tetrameric pores (2). But it has also been found to enable passage of larger molecules, including macromolecules under other experimental conditions (3). At low hydration and high concentration, melittin spans some bilayers at equilibrium, consistent with a toroidal pore structure (4, 5). But under most other experimental conditions, melittin has been shown by Fourier-transform infrared spectroscopy (4), oriented circular dichroism (6), x-ray diffraction (6), and electron paramagnetic resonance (7) to have its helical axis predominantly parallel to the membrane surface at equilibrium. Furthermore, electrochemical impedance spectroscopy (8) and dye leakage studies (9) have shown that melittin forms only transient, not equilibrium, pathways through synthetic bilayers that exist only for a brief time after the peptide first encounters the membrane surface. Shortly after leakage begins, it slows or stops completely (see Fig. 1).

Figure 1.

Transient pore formation by melittin in PC bilayers. (Red curve) Resistance of a supported bilayer after addition of melittin at a peptide/lipid ratio of 1:200 shows a rapid permeabilization followed by recovery of bilayer barrier properties (8). (Black curve) Ensemble rate of release of the dye quencher pair ANTS/DPX from lipid vesicles made from PC lipids shows a rapid permeabilization followed by a cessation of leakage before all contents have been released. In this experiment, peptide/lipid is also 1:200. (Inset) Given here is the leakage data from which the black curve was determined. To see this figure in color, go online.

In this issue, Weisshaar and colleagues (10) provide an intriguing and novel view of the mechanism of membrane permeabilization by melittin. They use fluorescence microscopy to observe, spatially and temporally, the effects of melittin on both the outer membrane (OM) and inner, cytoplasmic membrane (CM) of individual Escherichia coli cells. Although not an antimicrobial peptide by Nature’s design, melittin has potent bactericidal activity that mimics natural examples. This study reveals a surprisingly complex series of events on a kinetic path between initial exposure of bacteria to melittin and the ultimate loss of membrane barrier properties. These events provide clues to a broader understanding of the mechanistic landscape of melittin and the possible landscapes of the many other antimicrobial and membrane-permeabilizing peptides.

This article addresses how melittin acts on two real biological membranes. Specifically, the researchers observed fluorescence in individual Escherichia coli cells containing green fluorescent protein (GFP) that had been directed to the periplasmic space between the two membranes, enabling the independent detection of OM and CM permeabilization. They also monitored SYTOX Orange, a nonmembrane-permeant dye that becomes fluorescent only upon crossing a leaky CM and binding to DNA. They monitored fluorescence as a melittin solution of 10 μM, twice the lethal concentration, which was flowed over immobilized, live bacteria. The highly cationic, amphipathic peptide will rapidly accumulate on the bacterial LPS and OM. Yet, in the microscope, nothing appears to happen for minutes to tens of minutes. Then, over the course of 30 s or so, a rapid sequence of events is stochastically initiated in individual bacteria. First, the outer membrane suddenly becomes permeable to periplasmic GFP and the cells shrink rapidly. Presumably melittin now accumulates on the inner cytoplasmic membrane. Perhaps due to the rapid cell shrinkage, inward-facing membrane blebs (i.e., bubbles) transiently appear in the CM. The bubble membranes rapidly become permeable to SYTOX, and seconds later, to GFP, which flows inward from the periplasm through the bubble membrane.

Incredibly, 10–20 s after the sudden OM permeabilization event, both membranes reseal! Despite the continued presence of enough melittin to have made macromolecule-sized breaches in the integrity of both membranes, they sequentially become impermeant to GFP. The resealed state is very long lived, relative to the just described transient membrane permeabilization events, lasting for minutes to tens of minutes. Eventually, the bacteria develop a low, but constant level of membrane leakage that continues until the cells are destroyed.

The successive permeabilization of the two bacterial membranes to macromolecules, followed by successive resealing, supports an idea that has long existed on the fringes of the field: permeabilization of membranes, under some conditions, is mechanistically coupled to the dissipation of peptide asymmetry by sudden, cooperative peptide translocation events (11). Thus, permeabilization is transient and stochastic on the scale of single vesicles or single bacteria. In this kinetic variant of the classical carpet model (12), peptide accumulates on the outer monolayer of a bilayer, creating a growing, energetically unfavorable asymmetry of mass, charge, and surface pressure. The asymmetry is dissipated by a local, short-lived, catastrophic reorganization of the bilayer that allows bound peptide and other polar molecules to cross the bilayer and establish similar concentrations of peptide on both sides. Permeation ceases or slows once the peptide asymmetry has dissipated, presumably because the peptide at equilibrium is far less disruptive than the same peptide when asymmetrically distributed. Melittin, acting on PC vesicles (8, 9) or supported PC bilayers (8), indeed shows such transient permeabilization and resealing (Fig. 1), as do many other membrane-permeabilizing peptides. Thanks to Weisshaar and colleagues, we now know that melittin, acting on bacterial membranes, may behave this way also.

Binary mechanistic questions about melittin and other membrane-active peptides have dominated the field for decades (e.g., Does the peptide form small channels or large defects in the membrane? Does it form equilibrium pores or transient pores? Does it act like a detergent or like a membrane protein?) implying that there is a single most correct answer. But such isolated binary queries may be the wrong ones to ask about peptides in membranes because different mechanisms predominate under different experimental conditions, as exemplified by the behavior of melittin in synthetic bilayers (Fig. 1) and E. coli membranes (10). After 50 years of asking and answering (and arguing about) such limiting questions, perhaps it is time to more consistently discuss peptides in membranes in terms of mechanistic landscapes. In this way, the mechanism is strongly influenced by experimental details, and sets of experimental variables that provoke certain mechanistic behaviors comprise the useful experimental observations.

Editor: Claudia Steinem.