The Gram-Negative Outer Membrane: Structure, Biochemistry and Vaccine Potential Peter Owen Department of Microbiology, Moyne Institute, Trinity College, Dublin 2, Ireland.

Introduction

Royal Irish Academy Medal Lectuit-e

For Gram-negative bacteria, the region of the cell interfacing directly with the host/environment is the outer membrane. This membrane system is about 10 nm in diameter and is structurally, compositionally and functionally quite distinct from the cell’s cytoplasmic or inner membrane. Its main role is a protective one. It is composed of outer membrane proteins (OMPs; 44%), some phospholipids ( z 13%) and carbohydrate polymers ( % 43%), and is unusual in as much as it is a fully asymmetric bilayer. The inner monolayer is composed largely of phospholipid and the acyl chains of lipoproteins, whereas the outer monolayer contains lipopolysaccharide (LPS) [ 1,21. Outer membranes generally contain a restricted number (three to eight) of major OMPs present in high copy number (50 000-750 000 copiedcell). In the model Gram-negative bacterium Escherichia coli, several of these (e.g. Braun lipoprotein, peptidoglycan-associated lipoprotein, and the ompA gene product) appear to play a structural role in anchoring the outer membrane to the underlying peptidoglycan layer [3, 41. Another major and important class of proteins is characterized by the OmpF, OmpC, PhoE and LamB porins which facilitate the passive diffusion of small hydrophilic solutes through the outer membrane [ S , 61. Other proteins, such as the iron-regulated OMPs, are involved in TonB-dependent uptake of more specific solutes [7, 81.

Delivered on I 3 September I99 I at St Patrick’s College, Maynooth

%

Abbreviations used: Ag43, antigen 43; CIE, crossed immunoelectrophoresis; Fm, formyl; LPS, lipopolysaccharide; OMP, outer membrane protein; PEV, polyvalent extract vaccine; Quin, 2-amino-2,6-dideoxyglucose (quinovosamine).

nPROFESSOR PETER OWEN

For some pathogenic bacteria selected OMPs may act as virulence determinants. Thus, some have been implicated in the ability of pathogens to acquire iron, in mediating adherance to epithelial

I992

Biochemical Society Transactions

2

ecule into the outer membrane, a core polysaccharide domain and an 0-antigen region. The 0-antigen domain is most distal from the cell surface and is a repeat of a short oligosaccharide. For any one cell, the number of 0-antigen repeats can vary, thus giving a spectrum of heterogeneously sized LPS molecules (Fig. 1c). Polysaccharide antigens (capsules and LPS) can be important virulence determinants. Indeed, many pathogens utilize them to thwart and evade the immune system of the host by mechanisms which often relate to their anti-phagocytic properties, their resistance to opsonization, and to molecular mimicry [30]. In some instances, these polymers provide the basis of efficacious vaccines affording lasting protection to infection e.g. Type b capsule of H. influenzue, the A, C, Y and W135 capsules of h! rneningithiis, the capsules (23 serotypes) of Streptococcus pneumoniae and the 0-antigen (9 serotypes) of Ps. ueruginosu [ 15,27, 31-36]. The remainder of this article will be devoted to a description of two components of the bacterial outer membrane which havelmay have vaccinogenic potential viz the 0-antigen of Ps. uemginosu and a novel OMP of E. coli termed antigen 43 (4743).

cells, in the invasion of host cells, and also in thwarting the immune system [9, 101. Of more immediate relevance, however, is the fact that some OMPs have significant potential as vaccine constituents [ 111. Notable in this respect are the pertactins from Bordetella pertussis [ 12- 151 and B. bronchisepticu [16, 171 (the causative agents of whooping cough in humans and atrophic rhinitislpneumonia in pigs), Class 1 and Class 2 OMPs from NeisSeria meningitih [18, 191 (a leading cause of bacterial meningitis), Protein 1 from the gonococcus [ZO, 211, Protein F of Pseudomonus uemginosu [22-241, and 98000-M, and 46000-M, proteins from Huemophilus injluemue Type b [25, 261 (another major cause of bacterial meningitis). It should perhaps be emphasized that several of these vaccines are at an early experimental stage. Protein and phospholipid apart, the other main constituent of the Gram-negative outer membrane is carbohydrate. In E. coli, this takes several forms viz LPS, a polymer called enterobacterial common antigen and, if the strain is encapsulated, capsullar polysaccharide. These are anchored on the outer leaflet of the outer membrane and, apart from surface appendages such as flagella and fimbriae, represent the major antigens expressed on the bacterial surface [27,28]. Capsullar polysaccharides are relatively simple molecules which usually take the form of either linear homopolysaccharides or repeating heteropolysaccharides (see Fig. l a & lb). LPS, on the other hand, is a much more complex polymer [29]. For the purposes of this article it is sufficient to note that LPS is composed of three discrete areas viz a toxic Lipid A region which anchors the mol-

General methodology The methodologies employed cross a number of disciplines. At the analytical level, extensive use is made of high resolution immunochemical methods such as crossed immunoelectrophoresis (CIE) for studying membrane-associated antigens in undenatured form. A common experimental approach involves (a), resolution of membrane antigens by

Fig. I Structures of ( a ) , the homopolysaccharide capsule of E. coli K I ; ( b ) , the heteropolysaccharide capsule of E. coli K5; (c), LPS; and ( d ) , the 0-antigen of Ps. aeruginosa serotype 6 (Fisher I ) In ( a ) not all sugar residues are 0-acetylated and there are some internal ester bridges. In (c), n is variable, and structural detail has been omitted for simplicity. Data from (28, 451.

(4

-E)-a-NeuAc-(l-

0-Ac -4)-~-GlcNAc-( I -4)-B-GkA( I

-

(b)

[0-Antigen], - - -[Core polysaccharide] - - -[Lipid A]

-

-4)-a-~-GalNAcA-( I -4)-a-~-GalNFmA-( I 3)-a-~-QuiNAc-( I +2)-a-~-Rha-( I 3 6 6 t t t OAc NH2 NH2

Volume 20

-

(4 (d)

The Gram-Negative Outer Membrane

CIE; (b), identification of the component(s) of interest by zymograms, analyses of specific cofactors and polypeptides, reactions with defined antisera etc.; (c), production of monospecific antiserum by procedures involving precipitate excision; and (4, immunochemical, biochemical and biophysical analysis of the (immuno)purified antigen [371.

0-antigen of Ps. aeruginosa Ps. aeruginosa is an opportunistic and common nosocomial pathogen. It can be a serious problem for certain patient groups notably the immunocompromized and those suffering from severe burns or cystic fibrosis. The decrease in the welfare of patients succumbing to pseudomonal septicaemia is relatively rapid, and the mortality rate correspondingly high. Hence there is interest in developing a vaccine capable of inducing high levels of protective (opsonizing) antibody directed against pathogenic strains of this organism [38]. A crude vaccine termed polyvalent extract vaccine (PEV; Wellcome Biotech Ltd, Beckenham, Kent, U.K:) has been developed among others [39]. Encouraging results from field trials using PEV [40-421 indicated that active and passive immunization therapy could substantially reduce mortality caused by Ps. aeruginosa in burns patients. To prepare PEV, washed cells of individual serotypes are incubated in a glycine/EDTA buffer. Cells, which remain fully viable during this procedure, are then removed by centrifugation. The supernatant fraction, which is essentially a surface wash, is filtered to yield the monovalent extract. This extract is then combined with similar extracts prepared from 15 other serotype strains of Ps. aeruginosa to give the final product, PEV [39]. The following paragraphs outline experimentation defining the identity of the protective antigens in PEV. Notable features of analyses of PEV and of individual monovalent extracts include the following. First, PEV is resolved into ten major and several minor antigens when analysed by CIE against homologous anti-PEV serum. In contrast, each of the monovalent extracts reveals the presence of one or at most two major antigens when analysed under the same conditions. It can be shown that each extract contributes a major antigen to PEV. Secondly, the form of the CIE immunoprecipitates resolved for PEV is characteristic of carbohydrate immunogens. Thirdly, all monovalent extracts and PEV show substantial levels of ketodeoxyoctonate (a marker molecule for LPS) and reveal a laddet profile highly characteristic of heterogeneously sized smooth LPS (LPS molecules

with differing numbers of 0-antigen side-chains) when analysed by SDSIPAGE and silver staining for carbohydrate. By purifying the LPS from the 16 different serotype strains it can be demonstrated unequivocally that the major antigen resolved for each monovalent extract by CIE is LPS from the corresponding serotype [38,43]. Several minor protein antigens are also detected in PEV and in its component monovalent extracts. The question naturally arises as to whether LPS or these other minor components are responsible for the protective capacity of the vaccine. This issue has been resolved in a series of comprehensive protection experiments performed in mice using LPS, monovalent extracts, and monovalent extracts from which minor antigens had been selectively removed - all at equivalent LPS doses. Results indicate that there are sufficient molecules of LPS in each preparation to account for its full protective capacity - convincing evidence that LPS is the major protective antigen in PEV [44]. The above conclusion can be confirmed and extended in protection experiments performed on LPS preparations subfractionated by gel filtration into molecules containing differing numbers of 0-antigen repeat units. Thus, it has been shown that molecules containing > 18 0-antigen repeats are 50-100-fold more protective than ones bearing predominantly a single repeat, and that over 85% of the protection afforded by LPS could be accounted for by molecules with 10 or more repeating units. In addition, it can be calculated that as few as 2 x 10" of these molecules of LPS will confer 50% protection to mice against one 95% lethal dose of virulent Ps.aeruginosa [441. Thus, taking the specific example of monovalent extract 6, it is possible to state with some confidence that the protective antigens are LPS subspecies containing over 10 repeats of a structurally defined [45] linear tetrasaccharide. This contains 2-acetamido-2-deoxy-3-O-acetyl-n-galacturonamide, Z-formamido-Z-deoxy-i)-galacturonamide,2acetamido-2,6-dideoxy-i~-glucopyranose and L-rhamnose (Fig. 1s). These and other data give relevance to a growing body of evidence which suggests that intravenous transfer of anti-LPS immunoglobulins may be a useful, although not necessarily suficient, strategy in the control of Pseudomonas infection in the immunocompromized and critically ill patient. Furthermore, the ability to pinpoint the precise chemical nature of the protective antigen in a crude vaccine of proven efficacy provides a solid foundation on which to devise an improved and more

I992

3

Biochemical Society Transactions

acceptable vaccine based on a defined number of purified constituents [38].

Ag43 of E. coli 4

Ag43 was discovered during a comprehensive CIE analysis of outer membrane antigens in E. coli [46]. Clear resolution of Ag43 facilitated precipitate excision, production of quality anti-43 antiserum [43], and subsequent immunochemical and biochemical analysis of the immunopurified antigen [ 101. Ag43 can be present in copy numbers up to = 50000/cell and is composed of two dissimilar subunits (termed a'j and B'3) in 1:1 stoichiometry. a'3 has an M, of 60000. B'j (M,=53000) is heat modifiable and displays an apparent M, of 37000 when solubilized at T < 70°C. The two subunits are chemically and immunologically distinct and are not linked by disulphide bonds. Neither appears to be associated with the peptidoglycan nor to contain detectable LPS, enzyme activity, fatty-acyl groups or other cofactors [47]. The presence in the outer membrane of an antigen of this type is interesting and unusual, since most OMPs exist either as a single polypeptide chain or as homopolymers thereof (usually trimers). Further experiments using antiserum specific for only a'3 or B" reveals that the association between the two subunits is not artifactual. Thus, whereas it can be shown by Western blotting that anti-a4' and anti-B4j sera are subunit specific, both antisera can precipitate both subunits from detergent-solubilized, non-denatured memcomplex can branes [47]. Furthermore, the a43-/343 be reconstituted from mixtures containing purified a J 3and a-stripped membranes bearing only the /?" subunit [48].Cross-linking experiments performed with dithiobis(succinimidy1propionate) additionally indicate a '-a '' interactions and associations between a" and an 80000-M, protein, possibly the FepA protein [M. Meehan & P. Owen, unpublished work]. a4j but not /?" can be selectively released from outer membranes following brief heating to 60°C. This is a property common to several bacterial surface appendages such as fimbriae. In addition, the N-terminus of purified a43 reveals a sequence XXTVNGGTXXXXGXX present in the N-termini of certain enterobacterial fimbrial subunits [48]. However, is not in the same M, range as that of most fimbrial subunits ( M r , 18000-25 000) and electron microscopic examination indicates that Ag43 does not form an immediately recognizable surface structure. However, like fimbriae, a4j is highly immunogenic and is expressed on the surface of wild-type cells bearing

Volume 20

smooth LPS. This can be shown quite readily by immunoadsorption experiments conducted with whole cells [47], by electron microscopy performed in conjunction with immunogold labelling, and by fluorescent labelling. In contrast, /I"is only accessible to antibody in strains possessing rough LPS [P. Caffrey, J. Reesley, M. Meehan & P. Owen, unpublished work]. Based on the above and other evidence the following model can be proposed for the organization of Ag43 in the outer membrane. The aJ3subunit is essentially a peripheral protein expressed on the outer surface of the outer membrane and anchored to it by virtue of specific protein-protein interactions with /343, itself an integral outer-membrane protein. a4j penetrates the 0-antigen chains of LPS and is a major surface antigen. 0'' resides closer to the membrane surface, and its determinants only become unmasked in strains bearing truncated (rough) LPS molecules. Another interesting aspect of Ag43 is its ability to undergo reversible phase variation with a frequency (on-off) of about 4 x 10-j. This can be clearly shown by colony immunoblotting and by fluorescent microscopy using anti-43 immunoglobulins [ 481. Phase and (antigenic) variation are not novel properties for bacterial surface antigens. They have been described for certain fimbriae, flagella, capsules, and LPS [49]. However, this is probably the first documentation of the phenomenon for an OMP of E. coli [lo]. Ag43 seems to be species (EscheriChk) specific, but is expressed in a wide variety of E. coli strains including enteropathogenic isolates grown in nitro and uropathogenic strains harvested directly from infected urine [47, SO]. The precise function of Ag43 is unclear at the present time.. Initial studies do not support a role in adhesion. It may be a vestige of the highly ordered S-layers detected in many Gram-negative bacteria but reportedly absent from E. coli and related members of the Enterobacteriaceue [ 5 11. Principal interests in the protein are twofold. First, Ag43 may represent a novel class of surface antigen (adhesin or S-layer) with an important role to play in pathogenesis. Secondly, the properties of the subunit (copy number, M,, antigenicity, and surface expression) suggest that it may have vaccinogenic potential per se and/or be an ideal candidate for the genetic insertion of foreign epitopes. In this regard, it appears to display several favourable advantages over other candidate molecules such as the ompA gene product, the LamR protein, P-fimbriae and flagellin [ 521.

The Gram-Negative Outer Membrane

1. Nikaido. H. & Vaara, M. (1985) Microbiol. Rev. 49, 1-32 2. Hancock, R. E. W. (1991) ASM News 57,175-182 3. Lugtenberg, €3. & van Alphen. I,. (1983) Biochim. Biophys. Acta 737, 5 1- 1 15 4. Lugtenberg, B. (1985) in Enterobacterial Surface Antigens: Methods for Molecular Characterization (Korhonen. T. K., Dawes, E. A. & Makela, P. H., eds.), pp. 3- 16, Elsevier Science Publishers, Amsterdam 5. Benz, R. (1985) CRC Crit. Rev. Biochem. 19, 145- 190 6. Hancock, R. E. W. (1987) J. Bacteriol. 169,929-933 7. Postle, K. & Good, R. F. (1983) Proc. Natl. Acad. Sci. USA. 80,5235-5239 8. Postle, K. (1990) Mol. Microbiol. 4,2019-2025 9. Owen, P. (1988) in Immunochemical and Molecular Genetic Analysis of Bacterial Pathogens (Owen, P. & Foster, T. J.? eds.), pp. 27-43, Elsevier Science Publishers, Amsterdam 10. Owen, P. (1991) in Microbial Surface Components and Toxins in Relation to Pathogenesis (Ron, E. 2. & Rottem, S., eds.), pp. 127-139, Plenum Press, London 11. Gilleland, H. E., Jr. & Matthews-Greer, J. M. (1987) Eur. J. Clin. Microbiol. 6,231-233 12. Brennan, M. J., Li, 2. M., Cowell, J. L.,Bisher, M. E., Steven, A. C., Novotny, P. & Manclark, C. R. (1988) Infect. Immun. 56,3 189-3 195 13. Charles, I. G., Dougan, G., Pickard, D., Chatfield, S., Smith, M., Novotny, P., Morrissey, P. & Fairweather, N. F. (1989) Proc. Natl. Acad. Sci. USA. 86, 3554-3558 14. Shahin, R. D., Hrennan, M. J., Li, Z. M., Meade, B. D. & Manclark, C. R. (1990) J. Exp. Med. 171,63-73 15. Moxon, E. R. & Rappuoli, R. (1990) in Modern Vaccines: Current Practice and New Approaches (Moxon, E. R., ed.), pp. 100-112, Edward Arnold, London 16. Montaraz, J. A., Novotny, P. & Ivanyi, J. (1985) Infect. Immun. 47,744-75 1 17. Kobisch, M. & Novotny, P. (1990) Infect. Immun. 58, 352-357 18. Frasch. C. E., Tsai, C.-M. & Mocca, I,. F. (1986) Clin. Invest. Med. 9, 101-107 19. Frasch, C. E. (1987) Vaccine 5, 3-4 20. Jiskoot, W., Teerlink, T., van Hoof, M. M. M., Bartels, K., Kanhai. V., Crommelin, D. J. A. & Beuvery, E. C. (1986) Infect. Immun. 54,333-338 21. Jeurissen, S. H. M., Sminia, T. & Beuvery, E. C. (1987) Infect. Immun. 55,253-257 22. Gilleland, H. E., Jr., Parker, M. G., Matthews, J. M. & Berg, R. D. (1984) Infect. Immun. 44,49-54 23. Matthews-Greer, J. M. & Gilleland, H. E., Jr. (1987) J. Infect. Dis. 155, 1282-1291 24. Battershill, J. L., Speert, D. P. & Hancock, R. E. W. (1987) Infect. Immun. 55,2531-2533 25. Kimura, A., Gulig, P. A,, McCracken, G. H., Jr.,

Loftus, T. A. & Hansen, E. J. (1985) Infect. Immun. 47,253-259 26. Loeb, M. R. (1987) Infect. Immun. 55,2612-2618 27. Arbuthnott, J. P., Owen, P. & Russell, R. J. (1983) in Topley and Wilson’s Principles of Bacteriology, Virology and Immunity (Wilson, G. & Dick, H. M., eds.), vol. 1, pp. 337-373, Edward Arnold, London 28. Jann, K. & Jann, B. (1987) Rev. Infect. Dis. 9, (Suppl. 5), S517-S526 29. Kietschel, E. Th., ed. (1984) Handbook of Endotoxin vol. 1: Chemistry of Endotoxin, pp. 1-419, Elsevier Science Publishers, Amsterdam 30. Mims, C. A., ed. (1987) The Pathogenesis of Infectious Disease, pp. 1-342, Academic Press, London 31. Robbins, J. B., Schneerson, R. & Pittman, M. (1984) in Bacterial Vaccines (Germanier, R., ed.), pp. 289-3 16, Academic Press, London 32. Gotschlich, E. C. (1984) in Bacterial Vaccines (Germanier, R., ed.), pp. 237-255, Academic Press, London 33. Shann, F. (1990) in Modern Vaccines: Current Practice and New Approaches (Moxon, E. R., ed.), pp. 84-90, Edward Arnold, London 34. Austrian, R. (1984) in Bacterial Vaccines (Germanier, R., ed.), pp. 257-288, Academic Press, London 35. Cryz, S J., Jr. (1984) in Bacterial Vaccines (Germanier, R., ed.), pp. 3 17-35 1, Academic Press, London 36. Cryz, S. J., Jr., Sadoff, J. C., Furer, E. & Germanier, R. (1986) J. Infect. Dis. 154,682-688 37. Owen, I-’. (1985) in Enterobacterial Surface Antigens: Methods for Molecular Characterization (Korhonen, T. K. K., Dawes, E. A. & Makela, P. H., eds.), Ch. 15, pp. 207-242, Elsevier Science Publishers, Amsterdam 38. MacIntyre, S., McVeigh, T. & Owen, P. (1988) in Immunochemical and Molecular Genetic Analysis of Bacterial Pathogens (Owen, P. & Foster, T. J., eds.), Ch. 10, pp. 133-146, Elsevier Science Publishers, Amsterdam 39. Miler, J. J., Spilsbury, J. F., Jones, R. J., Roe, E. A. & Lowbury, E. J. I,. (1976) J. Med. Microbiol. 10, 19-27 40. Jones, R. J., Roe, E. A., Lowbury, E. J. I,., Miler, J. J. & Spilsbury, J. F. (1976) J. Hyg. 76,429-439 41. Jones, R. J. (1979) J. Hyg. 82,453-462 42. Roe, E. A. & Jones, R. J. (1983) Rev. Infect. Dis. 5, (Suppl. 5), S922-S930 43. MacIntyre, S., McVeigh, T. & Owen, P. (1986) Infect. Immun. 51,675-686 44. MacIntyre, S., Lucken, R. & Owen, P. (1986) Infect. Immun. 52,76-84 45. Knirel, Y. A. (1990) CRC Crit. Rev. Microbiol. 17, 273-304 46. Owen, P. (1983) in Electroimmunochemical Analysis of Membrane Proteins (Bjerrum, 0. J., ed.), Ch. 19, pp. 347-373, Elsevier Science Publishers, Amsterdam 47. Owen, P., Caffrey, P. & Josefsson, L.-G. (1987) J.

I992

5

Biochemical Society Transactions

6

Racteriol. 169, 3770-3777 48. Caffrey, P. & Owen, P. (1989) J. Racteriol. 171, 3634-3640 49. Birkbeck, T. H. & Penn, C. W., eds. (1986) Antigenic Variation in Infectious Diseases, pp. 1- 170, IRI, Press, Oxford 50. Meehan, M., Caffrey, P. & Owen. P. (1991) Abstr. 119th Ord. Meet. SOC.Gen. Microbiol. P8, 45

Volume 20

51. Smit, J. (1987) in Bacterial Outer Membranes as Model Systems (Inouye. M.. ed.), pp. 343-376, John Wiley & Sons, New York 52. Freudl, R., MacIntyre, S., Degen, M. & Henning, I T . (1986)J. Mol. Riol. 188,491-494 Received 28 August 1991