The crystal structure of the diphtheria toxin dimer at 2.5 A resolution reveals a Y-shaped molecule of three domains. The catalytic domain, called fragment A, is of the alpha + beta type. Fragment B actually consists of two domains. The transmembrane domain consists of nine alpha-helices, two pairs of which are unusually apolar and may participate in pH-triggered membrane insertion and translocation. The receptor-binding domain is a flattened beta-barrel with a jelly-roll-like topology. Three distinct functions of the toxin, each carried out by a separate structural domain, can be useful in designing chimaeric proteins, such as immunotoxins, in which the receptor-binding domain is substituted with antibodies to target other cell types.

Mucosal surfaces provide first-line defense against microbial invasion through their complex secretions. The antimicrobial activities of proteins in these secretions have been well delineated, but the contributions of lipids to mucosal defense have not been defined. We found that normal human nasal fluid contains all major lipid classes (in micrograms per milliliter), as well as lipoproteins and apolipoprotein A-I. The predominant less polar lipids were myristic, palmitic, palmitoleic, stearic, oleic, and linoleic acid, cholesterol, and cholesteryl palmitate, cholesteryl linoleate, and cholesteryl arachidonate. Normal human bronchioepithelial cell secretions exhibited a similar lipid composition. Removal of less-polar lipids significantly decreased the inherent antibacterial activity of nasal fluid against Pseudomonas aeruginosa, which was in part restored after replenishing the lipids. Furthermore, lipids extracted from nasal fluid exerted direct antibacterial activity in synergism with the antimicrobial human neutrophil peptide HNP-2 and liposomal formulations of cholesteryl linoleate and cholesteryl arachidonate were active against P. aeruginosa at physiological concentrations as found in nasal fluid and exerted inhibitory activity against other Gram-negative and Gram-positive bacteria. These data suggest that host-derived lipids contribute to mucosal defense. The emerging concept of host-derived antimicrobial lipids unveils novel roads to a better understanding of the immunology of infectious diseases.

Defensins, a family of cationic peptides isolated from mammalian granulocytes and believed to permeabilize membranes, were tested for their ability to cause fusion and lysis of liposomes. Unlike a-helical peptides whose lytic effects have been extensively studied, the defensins consist primarily of 0-sheet. Defensins fuse and lyse negatively charged liposomes but display reduced activity with neutral liposomes. These and other experiments suggest that fusion and lysis is mediated primarily by electrostatic forces and to a lesser extent, by hydrophobic interactions.Circular dichroism and fluorescence spectroscopy of native defensins indicate that the amphiphilic 0-sheet structure is maintained throughout the fusion process. Taken together, these results support the idea that proteinmediated membrane fusion depends not only on hydrophobic and electrostatic forces but also on the spatial arrangement of the amino acid residues to form a three-dimensional amphiphilic structure, which promotes the efficient mixing of the lipids between membranes. A molecular model for membrane fusion by defensins is presented, which takes into account the contributions of electrostatic forces, hydrophobic interactions, and structural amphiphilicity.

The ability of amphotericin B (AmB) to form ion-permeable channels in cholesterol containing lipid bilayers was studied by UV/visible absorbance, circular dichroism, and fluorescence spectroscopy. Stable liposomes composed of distearoylphosphatidylcholine, cholesterol, distearoylphosphatidylglycerol, and AmB were prepared so that a wide range of AmB concentrations in the bilayer could be studied. Singular value decomposition analysis (Henry & Hofrichter, 1992) of the circular dichroism spectra of AmB at different AmB/lipid ratios suggests that AmB exists primarily in only two states in the bilayer, a "monomeric" state and an "aggregated" state. The transition from the "monomeric" to the "aggregated" state begins to occur at a critical concentration of 1 AmB per 1000 lipids in the membrane and coincides with the appearance of channel activity. The data support the recent theoretical conclusions of Weakliem et al. (1995) which predict that pore formation in the lipid bilayer will occur when the drug molecule concentration exceeds a critical value. At this critical concentration, it is calculated that a minimum number of 16 AmB molecules per liposome are required to observe channel activity. The results are consistent with the sterol-dependent AmB channel models proposed by de Kruijff and Demel (1974), Andreoli (1974), and Khutorsky (1992). To further elucidate the effects of sterol on AmB-mediated channel formation, liposomes were prepared with varying ratios of cholesterol and AmB. At cholesterol mole percentages greater than 1, channel activity was observed to occur at AmB concentrations just above the critical value. Previous reports show that cholesterol forms "tail-to-tail" dimers at mole percentages greater than 2 (Harris et al., 1995). This suggests that formation of the bilayer-spanning channels by AmB is initiated most efficiently when the tail-to-tail dimer of cholesterol is present. Although the structural nature of the AmB channel could not be unambiguously determined, these experiments provide further evidence in support of the widely held view that AmB's primary mechanism of killing fungal cells occurs by forming ion-permeable channels.