Pathogenic bacteria secrete pore-forming toxins (PFTs) to attack target cells. PFTs are expressed as water-soluble monomeric proteins, which oligomerize into nonlytic prepore intermediates on the target cell membrane before forming membrane-spanning pores. Despite a wealth of biochemical data, the lack of high-resolution prepore structural information has hampered understanding of the b-barrel formation process. Here, we report crystal structures of staphylococcal g-haemolysin and leucocidin prepores. The structures reveal a disordered bottom half of the b-barrel corresponding to the transmembrane region, and a rigid upper extramembrane half. Spectroscopic analysis of fluorescently labelled mutants confirmed that the prepore is distinct from the pore within the transmembrane region. Mutational analysis also indicates a pivotal role for the glycine residue located at the lipid-solvent interface as a 'joint' between the two halves of the b-barrel. These observations suggest a two-step transmembrane b-barrel pore formation mechanism in which the upper extramembrane and bottom transmembrane regions are formed independently.

Staphylococcus aureus bi-component pore-forming toxins consist of S- and F-components, and form hetero-octameric beta-barrel pores on target blood cell membranes. Among them, γ-haemolysin (Hlg2 and F-component of Luk (LukF)) and LukED (LukE and LukD) possess haemolytic activity, whereas the Panton-Valentine leukocidin (LukS-PV and LukF-PV) does not lyse human erythrocytes. Here, we focussed on four loop structures in the rim domain of S-component, namely loops -1, -2, -3 and -4, and found that replacement of Loop-4 in both Hlg2 and LukE with that of LukS-PV abolished their haemolytic activity. Furthermore, LukS-PV gained haemolytic activity by Loop-4 exchange with Hlg2 or LukE, suggesting that Loop-4 of these S-components determined erythrocyte specificity. LOOP-1 and -2 enhanced the erythrocytes-binding ability of both components. Although Hlg2 and LukE recognize Duffy antigen receptor for chemokines on human erythrocytes, the ability of Loop-4 was not complementary between Hlg2 and LukE. Exchange of Hlg2 with LukE Loop-4 showed weaker activity than intact Hlg2, and LukE mutant with Hlg2 Loop-4 lost its haemolytic activity in combination of LukD. Interestingly, the haemolytic activities of these Loop-4 exchange mutants were affected by F-component, namely LukF enhanced haemolytic activities of these Hlg2 and LukE Loop-4 mutants, and also haemolytic activity of LukS-PV mutant with LukE Loop-4.