Journal of Molecular Biology
Volume 276, Issue 2, 20 February 1998, Pages 325-330
Journal home page for Journal of Molecular Biology

Communication
Staphylococcal α-hemolysin can form hexamers in phospholipid bilayers1

https://doi.org/10.1006/jmbi.1997.1535Get rights and content

Abstract

Atomic force microscopy (AFM) was used to study the structure of the staphylococcal α-hemolysin (αHL) oligomer formed in supported phospholipid bilayers. In contrast to the recent X-ray crystallographic demonstration of a heptameric stoichiometry for the oligomer formed in deoxycholate (DOC) micelles, the high-resolution unprocessed AFM images unequivocally revealed a hexamer in these phospholipid bilayers. Independent support of this hexameric stoichiometry was obtained from the measurements of the lattice constant in the AFM images and from gel electrophoresis. Therefore, αHL can form two different, energetically stable oligomers, which differ in at least stoichiometry but perhaps subunit structure as well. Furthermore, stable, incomplete oligomers were observed in the AFM images, which may be of relevance to the mechanism by which αHL damages the cell.

Introduction

Staphylococcal α-hemolysin (αHL) is a 33.2 kDa, water-soluble protein that converts into a pore-forming homooligomer upon binding to the plasma membrane (Bhakdi & Tranum-Jensen, 1991). Like other cytolytic toxins (Bhakdi et al., 1996), oligomerization of αHL proceeds spontaneously upon contact with a suitable amphipathic substrate, inasmuch as apparently indistinguishable oligomers of αHL form in cell membranes, detergent deoxycholate (DOC) micelles, and pure phospholipid bilayers. For many years, the αHL oligomer was thought to be a hexamer based on electron microscopy (EM) and biochemical studies (for a summary, see Gouaux et al., 1994). However, recently X-ray crystallography (Song et al., 1996) clearly demonstrated that the αHL oligomer formed in DOC micelles is a heptamer, and since this demonstration the hexameric model has been largely disregarded Bhakdi et al 1996, Song et al 1996. It was argued (Gouaux et al., 1994) that the image processing techniques used in EM could have imposed an artifactual symmetry onto the αHL oligomer, leading to an erroneous assessment of its subunit stoichiometry, and that the biochemical assays may not have had the required resolution to make an unambiguous assignment. However, the outer diameter of the αHL oligomer determined with EM, which should be a more reliable measurement Ribi et al 1988, Frank et al 1995 than the determination of stoichiometry, was observed to be between 7 nm and 9 nm Olofsson et al 1988, Ward and Leonard 1992, significantly smaller than the 10 nm diameter obtained by X-ray crystallography of the oligomer formed in DOC micelles (Song et al., 1996). This difference hence cannot simply be attributed to the limited accuracy of EM.

Since atomic force microscopy (AFM) has accurately resolved, under nearly physiological conditions, the structures and subunit stoichiometries of many soluble Erie et al 1994, Fritz et al 1995, Mou et al 1996a, Mou et al 1996b, Muller et al 1997 and membrane proteins Mou et al 1995, Schabert et al 1994, Muller et al 1996, Walz et al 1996 without using averaging techniques, we have applied this new approach to re-evaluate the structure of the αHL oligomer in a membrane. We have found that αHL can indeed form hexamers in phospholipid bilayers. Furthermore, analysis of the AFM images suggests that the structure of the subunit in the hexamer could be significantly different from that in the heptamer. We have observed stable, incomplete oligomers in these AFM images, which may be relevant to the mechanism by which αHL damages a cell.

It has been established that in order to obtain high-resolution images with AFM, the molecules under study must be well adsorbed to a substrate and closely packed (Shao et al., 1996). Normally, this would require high concentrations of protein incubated for long periods of time (Fang et al., 1997), but these conditions were undesirable in this study because inactive oligomers of αHL can form spontaneously in solution at high concentrations (Arbuthnott et la., 1967) or from proteolyzed monomers, which accumulate over time (Blomqvist et al., 1987). Therefore, we developed a procedure, similar to that used in EM, using small wells in blocks of Teflon, and with this method, large supported membranes (greater than 10 μm), containing a high density of αHL oligomers, were reproducibly prepared using a relatively low concentration of toxin (less than 50 μg/ml) and with relatively short periods of incubation (less than 12 hours). As shown in Figure 1a, the hexagonally packed αHL oligomers are clearly detected by the AFM even at somewhat larger scan sizes, and at smaller scan sizes, the size, shape and subunit stoichiometry of the oligomer are directly resolved from the unprocessed images (Figure 1b and c). The piezoscanner used in these experiments was calibrated with mica, gold ruling and the cholera toxin B-oligomer; AFM images of the latter (Mouet al., 1995) were consistent with both X-ray and EM measurements Ribi et al 1988, Zhang et al 1995.

The AFM images show that each individual αHL oligomer in the phospholipid bilayer is unequivocally a hexamer (Figure 1b and c). This was observed reproducibly with different tips, different scan directions, and different samples under a variety of conditions, including in the presence of Ca2+ (1 mM CaCl2), which has been shown to block the pore but not the oligomerization Menestrina 1986, Thelestam et al 1991. The hexameric stoichiometry is further supported by the size of the lattice constant (7.6(±0.4) nm, n = 46) in the AFM images, which demonstrates that the maximal diameter of the oligomer is significantly smaller than that of the heptamer (10 nm) formed in DOC micelles (Song et al., 1996), but within the range observed with EM Olofsson et al 1988, Ward and Leonard 1992. Since αHL oligomers, whether formed in liposomes or micelles, are stable in SDS Bhakdi et al 1981, Valeva et al 1995, the sizes of the oligomers prepared under these conditions were directly compared using SDS-PAGE (Figure 2). The oligomers formed in DOC micelles migrated more slowly than the oligomers from phospholipid vesicles (eggPC to BPS, 9 : 1) with either DOC or Triton X-100 (the latter detergent does not induce oligomerization of the water-soluble monomers; Bhakdi et al., 1981), suggesting that the oligomers formed in DOC micelles are slightly larger. Since the oligomers formed in the bilayer and then solubilized with DOC showed the same size as those solubilized with Triton X-100, the oligomer, once formed in the bilayer, could not be converted into what forms in DOC micelles (see Figure 2, lanes B), indicating that oligomerization is not reversible and that the oligomers formed under either condition are energetically stable. So far, we have been unable to image the oligomer formed in DOC with AFM; these oligomers do not adsorb strongly to mica, which is a well-documented problem with many protein samples Karrasch et al 1993, Shao et al 1996, nor do these oligomers readily incorporate into a supported membrane at a sufficiently high density (where the residual detergent is detrimental to the preformed bilayer). However, since it is reasonable to assume that the oligomers formed in DOC micelles are the same as those studied by X-ray crystallography (Song et al., 1996), we conclude from our observations that αHL can form oligomers with two different stoichiometries; heptamers and hexamers. Consistent with this conclusion, we note that in a recent AFM study (Fang et al., 1997) where images of oligomers of a mutant αHL were presented, the size of the oligomer appeared to be heterogeneous, even though a heptameric stoichiometry was concluded based on the average of 18 oligomers. At the moment, it is not known whether both oligomers can co-exist in the cell membrane, and the pathological role of two different oligomers is not clear.

Although inspection of the surface profiles suggests that the subunits in the two oligomers could be arranged similarly, a simple geometric argument, whereby one simply removes one subunit from the heptamer and then closes the ring, predicts that the diameter of the hexamer should be ∼85% of the diameter of the heptamer, instead of the observed 76%. Whether this difference in diameter is due to different subunit structures in each oligomer remains to be determined.

The total height of the hexamer, directly measured from the surface of the mica to the top of the oligomer by AFM, is 9.7(±1.2) nm (n = 47), in close agreement with that depicted in the X-ray model (10 nm), although whether αHL makes direct contact with the mica could not be determined. The hexamer, however, protrudes only 5.0(±0.5) nm (n = 76) from the bilayer (Figure 3), compared with 7 nm estimated from the X-ray structure (Song et al., 1996). This was measured whether or not the samples were fixed with glutaraldehyde, which, combined with the agreement in height between the two oligomers, indicates that there was little compression of the hexamer during imaging, as was noted previously with gap junctions (Hoh et al., 1991). This suggests that the two oligomers could have substantially different structures: assuming that the subunits are structurally similar, this height would require the polar rim domain (as defined by Song et al., 1996) to be completely inserted into the bilayer and the end of the hydrophobic stem to protrude farther from the bilayer than was predicted by the X-ray model (Song et al., 1996).

Another intriguing observation in the AFM images is the existence of oligomeric defects, in which one or more subunits are absent from the oligomer (Figure 1c). These oligomeric defects, present despite the high surface concentration of complete oligomers, were stable even after hours of repeated scanning, indicating that these arc-shaped complexes were energetically stable in the bilayer. Moreover, the general morphology of these incomplete oligomers suggests that the structure of the subunits should be similar to that in the complete hexamer. This raises an interesting question of whether such incomplete oligomers could be a measurable fraction in vivo, where the toxin concentration could be much lower that the concentration used in these studies. If such incomplete oligomers remain competent in membrane insertion, there might be an additional route of membrane damage by αHL similar to that by other bacterial toxins, such as streptolysin-O and other members of the cholesterol-binding cytolysins Alouf and Geoffroy 1991, Tweten 1995, Bhakdi et al 1996, which are believed to be capable of forming a pore in the membrane even when the oligomeric ring formed by these toxins is not closed.

In conclusion, staphylococcal αHL was observed to form hexamers in phospholipid bilayers, as was believed until recently. Together with the heptamers observed by X-ray crystallography, these results indicate a polymorphism of the subunit stoichiometry in the αHL oligomer. At present, the exact molecular mechanism by which αHL forms the two different oligomers is not understood, and whether the amphipathic substrate played a significant role in the formation of the two different oligomeric stoichiometries remains to be demonstrated. A full understanding of the underlying principles for the formation of one stoichiometry over the other will undoubtedly provide a valuable insight into the still poorly understood process of the assembly of membrane complexes.

Section snippets

Acknowledgements

We thank X. Wu for technical advice and assistance, and M. Stowell, M. Wiener, A. V. Somlyo and A. P. Somlyo for their critical reading of the manuscript. This work is supported by grants from the National Institutes of Health (RO1-RR07720 and PO1-HL48807) and the US Army Research Office (DAAL03-92-G-0002).

References (33)

  • T. Walz et al.

    Surface topographies at subnanometer resolution reveal asymmetry and sidedness of aquaporin-1

    J. Mol. Biol.

    (1996)
  • R.J. Ward et al.

    The Staphylococcus aureus alpha-toxin channel complex and the effect of Ca+2 ions on its interaction with lipid layers

    J. Struct. Biol.

    (1992)
  • R. Zhang et al.

    The 2.4 Å crystal structure of cholera toxin B subunit pentamercholeragenoid

    J. Mol. Biol.

    (1995)
  • J.E. Alouf et al.

    The family of the antigenically related cholesterol binding cytolytic toxins

  • J.P. Arbuthnott et al.

    Physical states of staphylococcal alpha-toxin

    J. Bacteriol.

    (1967)
  • S. Bhakdi et al.

    Alpha-toxin of Staphylococcus aureus

    Microbiol. Rev.

    (1991)
  • Cited by (0)

    1

    Edited by W. Baumeister

    View full text