Staphylococcus aureus plays a major role as a bacterial pathogen in human medicine, causing diseases that range from superficial skin and wound to systemic nosocomial infections . The majority of S. aureus strains produces a toxin, a proteinaceous exotoxin whose hemolytic, dermonecrotic, and lethal properties have long been known (1-6). The toxin is secreted as a single- chained, nonglycosylated polypeptide with a M(r) of 3.4 x 10(4) (7, 8). The protein spontaneously binds to lipid monolayers and bilayers (9-14), producing functional transmembrane pores that have been sized to 1.5-2.0-nm diameters (15-18). The majority of pores formed at high toxin concentrations (20 mug/ml) is visible in the electron microscope as circularized rings with central pores of approximately 2 nm in diameter. The rings have been isolated, and molecular weight determinations indicate that they represent hexamers of the native toxin (7). We have proposed that transmembrane leakiness is due to embedment of these ring structures in the bilayer, with molecular flux occurring through the central channels (15, 19). Pore formation is dissectable into two steps (20, 21). Toxin monomers first bind to the bilayer without invoking bilayer leakiness . Membrane-bound monomers then laterally diffuse and associate to form non-covalently bonded oligomers that generate the pores. When toxin pores form in membranes of nucleated cells, they may elicit detrimental secondary effects by serving as nonphysiologic calcium channels, influx of this cation triggering diverse reactions, including release of potent lipid mediators originating from the arachidonate cascade (22-24). That alpha toxin represents an important factor of staphylococcal pathogenicity has been clearly established in several models of animal infections through the use of genetically engineered bacterial strains deleted of an active alpha toxin gene (25-27). Whether the toxin is pathogenetically relevant in human disease, however, is a matter of continuing debate. Doubts surrounding this issue originate from two main findings. First, whereas 60 percent hemolysis of washed rabbit erythrocytes is effected by approximately 75 ng/ml alpha toxin, approximately 100-fold concentrations are required to effect similar lysis of human cells (4-6, 13). The general consensus is that human cells display a natural resistance towards toxin attack. The reason for the wide inter-species variations in susceptibility towards alpha toxin is unknown but does not seem to be due to the presence or absence of high-affinity binding sites on the respective target cells (20, 21). Second, low-density lipoprotein (28) and neutralizing antibodies present in plasma of all healthy human individuals inactivate a substantial fraction of alpha toxin in vitro. These inactivating mechanisms presumably further raise the concentration threshold required for effective toxin attack, and it is most unlikely that such high toxin levels will ever be encountered during infections in the human organism. The aforegoing arguments rest on the validity of two general assumptions. First, the noted natural resistance of human erythrocytes to alpha toxin must be exhibited by other human cells. Second, toxin neutralization by plasma components, usually tested and quantified after their preincubation with toxin in vitro, must be similarly effective under natural conditions, and protection afforded by these components must not be restricted to specific cell species.