Sir, An important aspect of microbial pathogenesis relates to the interaction between the invading pathogen and the host at the molecular level. Information on the three-dimensional structures of the component proteins involved can provide valuable clues concerning the mechanisms that underlie these processes. To date, none of the class 5 outer membrane proteins from Neisseria meningitidis has been investigated at the structural level. In order to obtain an insight into the molecular basis for meningococcal adhesion and recognition of Opc antigen by antibody, we have acquired low-resolution structural data of reconstituted Opc by using electron microscopy of two-dimensional crystals combined with crystallographic image analysis. Here, we present new structural data on the integral membrane protein Opc, an adhesin and invasin from the Gram-negative pathogen Neisseria meningitidis. Opc is a member of the class 5 group of integral outer membrane proteins and is expressed at high levels by certain strains of N. meningitidis (Achtman et al., 1988, J Exp Med168: 507–525). It is highly immunogenic in humans, and antibodies against Opc are bactericidal (Rosenqvist et al., 1993, J Infect Dis167: 1065–1073). There is also a substantial body of evidence indicating that Opc facilitates the adhesion and invasion of epithelial and endothelial cells by the recognition of specific cell surface receptors (Virji et al., 1992, Mol Microbiol6: 2785–2795;Virji et al., 1994, Mol Microbiol14: 173–184; de Vries et al., 1998, Mol Microbiol27: 1203–1212). A topological model has been proposed for Opc (Merker et al., 1997, Mol Microbiol23: 281–293), consisting of 10 transmembrane β-strands connected by five large surface-exposed loop regions. The model was based on epitope mapping, mutagenesis and insertion of a foreign epitope into the surface loops. It predicts that a substantial proportion of the total mass of Opc would protrude above the membrane, presumably to facilitate interaction with cell surface receptors in the host. Opc was purified from N. meningitidis as described previously (Achtman et al., 1988, J Exp Med168: 507–525). The protein was solubilized at a concentration of 5 mg ml−1 in 1% (v/v) C8E5 detergent (P-1115; Bachem) and combined with 1 mg ml−1 dimyristoyl phosphatidyl choline (DMPC), solubilized in 2% C8E5, to give a final ratio of lipid–protein of 0.1 (w/w). Opc was reconstituted into lipid vesicles by microdialysis to remove the C8E5 detergent over several days at 20°C against 25 mM sodium acetate/acetic acid (pH 4.5) and 50 mM ZnCl2. The sample was analysed by transmission electron microscopy using a uranyl acetate negative stain. Very little soluble protein was observed, indicating that the protein reconstituted with high efficiency. Vesicles measuring 1–2 μm in diameter were abundant, and small areas of ordered protein molecules were found consistently in approximately 5% of the vesicle population: an example of such an area is shown in Fig. 1A. . A. An area of a reconstituted Opc vesicle containing two-dimensional crystals. B. A Fast Fourier Transform of the two-dimensional crystal area shown in (A). C. An averaged p1 projection map of the merged crystal data (see Table 1). The two distinct circled domains are discussed in the main body of the text. Selected areas containing two-dimensional crystals were digitized and subjected to interactive image analysis using the program suite crisp (Hovmöller, 1992, Ultramicroscopy41: 121–135; Table 1). A computerized Fourier Transform of a typical area is shown in Fig. 1B: strong reflections are readily apparent at first-order lattice points, and weaker reflections were also present in FFTs from some crystals at reciprocal lattice points out to (h,k) = (3,1). By selecting data from several independent crystals and then merging together separate areas with similar unit cell parameters by phase origin refinement, a low-resolution projection map could be created based on the accumulated data (Henderson et al., 1986, Ultramicroscopy19: 147–156). The accuracy of the merging was monitored using the average interimage phase residual as an indication of the success of the overall merging procedure (Stoylova et al., 1997, Micron28: 439–446). The presence of a consistent regular lattice within the computed FFTs showed that Opc adopted one particular crystal form, with an average unit cell a = 67.3 ± 2.3 Å, b = 62.8 ± 2.1 Å, γ = 88.2 ± 2.8°. The correct orientation of each image with respect to the reference image was verified by inspection of the interimage phase residual: in every case, only one out of the two possible orientations (related by 180°) was clearly correct. In order to ensure that the final averaged structure did not result from the inclusion of spurious data, the merged data file was divided into two subsets, and each set was merged independently (Stoylova et al., 1998, Micron29: 341–348). The two resulting data sets produced projection maps that were in good agreement with the total merged data set. A projection map was calculated from the averaged Opc data, indexing the reflections in spacegroup p1 (i.e. without the imposition of any crystallographic symmetry). The map provides a view perpendicular to the plane of the membrane and shows two distinct main areas of staining, which are circled in Fig. 1C. The regions that exclude the uranyl acetate stain correspond to portions of the Opc protein that protrude above the membrane lipid bilayer; the image therefore provides information on the extramembranous portions of Opc. Within each stained region in Fig. 1C, three peaks of density can be seen: a possible interpretation of the data is that each of these peaks corresponds to a single Opc monomer and that a complete trimer is enclosed within each of the circled areas. A trimeric quaternary organization of Opc would be consistent with previous molecular weight measurements (Achtman et al., 1988, J Exp Med168: 507–525). A definitive assignment of quaternary structure and crystal packing should await higher resolution data, however. Assuming a trimeric organization of Opc, the size of each Opc monomer within the trimer measures approximately 20 Å across in the projection map, a feature that presumably corresponds to the predicted extrinsic loop region (Merker et al., 1997, Mol Microbiol23: 281–293). If Opc does indeed form a 10-stranded β-barrel within the membrane, then the diameter of the barrel can be estimated to be about 20 Å, based on a comparison with the dimensions of the 16-stranded β-barrel porin crystal structures. As negative stain primarily reveals the surface relief of the Opc molecule, we can infer that the regions of Opc corresponding to the extrinsic loops are of approximately the same size as the transmembrane β-barrel region. This would tend to suggest that each Opc monomer is compact and cylindrical in shape. The projection map presented here provides evidence that Opc probably has a trimeric oligomeric organization, and that each Opc monomer contains large portions of the protein protruding above the surface of the membrane. This is consistent with the function of Opc as an adhesin in non-encapsulated meningococci and also with the recognition of epitopes within the protein by antibody. The first structural details of this important membrane protein have, therefore, provided strong supporting evidence for its biological functions. The authors thank the Wellcome Trust, BBSRC, Astra Charnwood, Proctor and Gamble Pharmaceuticals and Merck, Sharp and Dohme for funding. The authors also thank Dr Mark Rosenberg for helpful discussion and assistance. J.P.D. is a Fellow of the Lister Institute of Preventive Medicine.
No takes yet. Share an insight, caveat, or question.
Collins et al. (1999) studied this question.