The importance of antibodies to LPS is suggested by the fact that innate “natural” antibodies to LPS develop in the absence of antigenic exposure (3
) and their passive transfer to antibody-deficient mice protects from lethal endotoxin challenge (45
). However, most anti-LPS antibodies in normal human sera have likely developed after exposure to antigen, either from pathogenic or commensal organisms (“naturally induced” antibodies) (14
) or as a result of immunization (2
The abundance and consistent expression of LPS in the outer membrane suggests that LPS should be a good target for bactericidal antibodies and a potential vaccine component for N. meningitidis
. But its toxicity and potential cross-reactivity with oligosaccharides on human cells have limited interest in it as a vaccine. Since the polysaccharide or oligosaccharide epitopes on LPS behave as T-cell-independent antigens, LPS would be expected to require association with protein(s) to be effective as a vaccine for infants (17
). Consequently, candidate vaccines for group B meningococcal disease and other diseases that are based on LPS or contain LPS as an important antigen have been designed to circumvent this problem by presenting the LPS as a covalent conjugate, as noncovalent complexes with proteins, or as part of an intact outer membrane (2
The effectiveness of conjugating polysaccharides to protein to convert them to T-cell-dependent antigens is well established. However, T-cell-independent antigens, including LPS, when presented in the context of whole bacteria or isolated noncovalent complexes with protein also appear to exhibit T-cell-dependent properties (5
). In any case, vaccines that induce anti-LPS antibodies have shown promise for inducing protection against diverse Gram-negative bacteria (5
). While efforts to develop an LPS-based vaccine for group B N. meningitidis
have been reported (10
), no vaccine candidate based solely on LPS has entered clinical trials. Some LPS-based vaccine efforts have focused on the inner core structures (10
) or a partially truncated LPS (56
) to ensure safety or improve immunogenicity (31
). Deoxycholate-extracted vesicles, which have been used for vaccination to respond to specific group B epidemics, have normally contained 5 to 7% residual LPS (14
), usually of the L3,7 immunotype.
Vaccines based on deoxycholate-extracted outer membrane vesicles typically induce bactericidal antibodies that are predominantly specific for PorA and are thus mostly subtype specific. A contribution of anti-LPS antibodies to the serum bactericidal activity of individuals vaccinated with deoxycholate-extracted outer membrane vesicle vaccines has been reported (54
), but the extent of the contribution is unclear, as is its contribution to the induction of cross-reactive bactericidal antibodies by those vaccines (48
). Part of the reason for the lack of understanding is the difficulty in determining the specificity of the bactericidal antibodies induced by complex vaccines, such as those based on outer membrane vesicles and by natural infections. By means of a bactericidal depletion assay (66
) developed in our laboratory, we have been able to assess the contribution of anti-LPS antibodies to the bactericidal activity of a variety of human sera. This approach requires only a small amount of serum and does not require the affinity purification of the antibodies in order to directly test them for bactericidal activity.
Koeberling et al. (30
) have shown the induction of anti-factor binding protein (fHBP) bactericidal antibodies after immunization of mice with NOMV, especially when expressed at levels 10-fold greater than the parent strain and compared to ΔfHBP NOMV. Higher bactericidal activity was detected against 5 strains expressing antigenically different porA
compared to the adjuvant-only control. Their method of adsorbing “bulk” quantities of sera immunized with 10× fHBP NOMV over a recombinant fHBP column and reporting a significant decrease in the bactericidal titer against several test strains is evidence for a strong anti-fHBP response. They did not detect a significant anti-LOS bactericidal response under those conditions, which were designed specifically to induce a strong anti-fHBP response. The reason Koeberling et al. did not see a significant bactericidal response to LPS is unclear. Such antibodies may have been difficult to detect in the presence of high titers of anti-fHBP bactericidal antibodies. They used a different bactericidal assay method which includes growth of the test strain in the presence of CMP-NANA to strongly sialylate the LPS. They also used different analytical methods to determine the specificity of the bactericidal antibodies, and in addition, 3 of 6 of their test strains expressed L1 LPS, which we have found to be poorly cross-reactive with L3,7 antibodies. It was not clear to what extent the LPS in their NOMV vaccines was sialylated. We have observed that sialylated LPS is less immunogenic than unsialylated LPS. Further work will be required to better understand the reasons for the different results.
In order to specifically identify broadly cross-reactive protective antigens, the bactericidal depletion assays summarized in this paper tested mostly human sera against strains heterologous to the strain that induced the bactericidal antibodies (e.g., convalescent-phase serum with the available case isolate). In other cases, such as normal human sera, the strain or strains that induced the bactericidal antibodies were unknown. Thus, we were less likely to detect strain- or type-specific bactericidal antibodies. For example, convalescent-phase sera from patients infected with group C organisms were, with a few exceptions, tested against group B test strains with serotype and serosubtype different from the causative strain. Nevertheless, titers against the two strains (6275 and B16B6) that shared the same PorA protein as the outbreak strain were not higher than those against PorA heterologous strains. Strain B16B6 was killed only by serum from the patient whose infecting strain expressed L2. The reason bactericidal antibodies to PorA were not present at titers comparable to the anti-LPS antibody titers in the convalescent-phase sera from the group C cases is not clear. PorA present on whole cells or in NOMV vaccines is likely less exposed than that present in deoxycholate-extracted vesicles from which much of the LPS, lipoprotein, and phospholipid have been removed. In NOMV vaccines which retain all the membrane-associated proteins, PorA is a significantly smaller percentage of the total protein. Thus, a dose of NOMV vaccine would contain a smaller amount of PorA than the deoxycholate-extracted vesicles. While some of the sera may have contained antibodies to the group B capsule, these antibodies are generally not bactericidal in conjunction with the human complement (63
). In the human sera tested, antibodies to LPS appeared to play a major role in the cross-reactive bactericidal activity observed. As might be expected, the LPS immunotype homologous to that expressed by the test strain was always the most efficient in depleting the bactericidal activity against that strain but was not the only immunotype that depleted bactericidal activity. There are likely different populations of anti-LPS antibodies in the sera that have different specificities and thus different avidities for the LPS expressed by a given test strain. The presence of bactericidal antibodies with L3,7 specificity in pooled normal human sera was also demonstrated by Estabrook et al. by the affinity purification method (13
It is not known to what extent the LPS on meningococci growing in vivo
in a host is sialylated, but the presence of endogenous sialyltransferases suggests there may be a high degree of sialylation (35
). If LPS is highly sialylated in vivo
, then protective LPS antibodies must be able to bind to sialylated LPS. Yet the failure of sialylated LNnT versus unsialylated LNnT to remove bactericidal antibodies suggests that the L3,7-specific bactericidal antibodies to the LNnT epitope recognize mainly unsialylated (L7) LPS. This specificity is similar to monoclonal antibody (1B2-1B7) (), known to be specific for LNnT, which was not able to kill a highly sialylated strain. Nonetheless, anti-LPS antibodies in most of the human sera were bactericidal for highly sialylated meningococci, suggesting that the human antibodies are directed mainly at epitopes not including sialic acid, likely involving some core structure as well as LNnT sugars. Studies with monoclonal antibody 9-2-L379 () have demonstrated a requirement for both the LNnT group and the L3 core structure for optimal binding (W. D. Zollinger, E. E. Moran, and B. L. Brandt, unpublished data). The bactericidal titer of this antibody was reduced but not eliminated by high sialylation of the test strain. In addition, if a subset of LPS-specific antibodies in human sera is fully directed to core structures, then it has the potential to be specific for unique LPS epitopes and yet bind to full-length, sialylated LPS on the meningococcus (see Fig. S2 in the supplemental material). Although most of the N. meningitidis
strains used as bactericidal test strains expressed full-length LPS containing LNnT, often L3,7 LPS, we observed that sera with bactericidal activity toward such strains were also able to kill strains expressing shorter L8 or L8-like LPS. The observation that human sera are able to kill strains expressing short and full-length LPS suggests the presence of antibodies that recognize primarily the core structure of the LPS and are not dependent on the presence of the LNnT group. The LPS-specific bactericidal activity of most of the human sera we tested were reduced by 2- to 4-fold by high sialylation of the test strain resulting from growth on CMP-NANA, but we could not determine the relative contributions of LPS sialylation and a more general increase in serum resistance to the decreased titer.
Induction of antibodies able to bind LPS with similar cores yet different alpha chain lengths after vaccination with an LPS-based vaccine would seem to be the most desirable from a safety point of view. The monoclonal antibody 21-1-LC1 (49
) appears to have these characteristics (see Fig. S2 in the supplemental material) in that it was made against an L5 strain expressing mostly the truncated L8-like form of the LPS (L8-5), as judged by silver-stained SDS-PAGE. The antibody appears to bind equally well on Western blots to full-length (including the sialylated form) and truncated L8-5 LPS. Vaccine-induced antibodies with these characteristics may be both safe and effective for protection against meningococcal disease. In the design of a multivalent group B vaccine based on NOMV, we have included the truncated L8 or L8-like LPS structures in the vaccine and have induced antibodies bactericidal for strains expressing full-length LPS (28
Some of the bactericidal antibody depletion data suggest that anti-LPS antibodies and anti-OMP antibodies may be able to cooperate in producing a bactericidal event (62
). This type of cooperative event would not likely be detected when looking at the bactericidal activity of affinity-purified anti-LPS antibodies (13
) or by simply trying to correlate binding on Western blots with bactericidal activity (54
). Frequently, two different purified antigens, such as LPS and recombinant protein, were used in bactericidal depletion assays, and the total amount of bactericidal activity removed by the two antigens added up to greater than 100%. These results suggest cooperation between antibodies specific for two different antigens in the killing event. Complement deposition induced by bactericidal antibodies to the two different antigens acting together would require that the antigens are located close together on the bacterial surface. If antibodies of two different specificities are involved, the removal of either of these antibody specificities could potentially result in loss of bactericidal activity. The cooperative (even synergistic) action of antibodies to two different OMPs has been reported (57
). Certainly, LPS is often closely associated with outer membrane proteins, so it would not be hard to imagine cooperation in killing between bactericidal antibodies specific for LPS and OMPs. Further work needs to be done to verify this hypothesis.
Our demonstration of the high prevalence of persistent anti-LPS bactericidal antibodies in normal, convalescent, and postvaccination human sera supports the use of appropriate forms of LPS as a vaccine or vaccine component.