Application of a Neisseria meningitidis antigen microarray to identify candidate vaccine proteins from a human Phase I clinical trial.

Meningococcal meningitis is a rare but serious condition affecting mainly children and young adults. Outer membrane vesicles (OMV) from Neisseria meningitidis have been used successfully as vaccines against the disease, although they only provide protection against a limited number of the many existing variants. There have been many attempts to identify suitable protein antigens for use in defined vaccines that provide broad protection against the disease, such as that leading to the development of the four component 4CMenB vaccine. We previously reported the use of a protein antigen microarray to screen for IgG antibodies in sera derived from human recipients of an OMV-based vaccine, as part of a Phase I clinical trial. Here, we show that computational methods can be used to cluster antigens that elicit similar responses in the same individuals. Fitting of IgG antibody binding data to 4,005 linear regressions identified pairs of antigens that exhibited significant correlations. Some were from the same antigens in different quaternary states, whilst others might be correlated for functional or immunological reasons. We also conducted statistical analyses to examine correlations between individual serum bactericidal antibody (SBA) titres and IgG reactivity against specific antigens. Both Kendall's tau and Spearman's rank correlation coefficient statistics identified specific antigens that correlated with log(SBA) titre in five different isolates. The principal antigens identified were PorA and PorB, RmpM, OpcA, and the type IV pilus assembly secretin, PilQ. Other minor antigens identified included a lipoprotein, two proteins from the BAM complex and the efflux channel MtrE. Our results suggest that consideration of the entire antigen composition, and allowance for potential interaction between antigens, could be valuable in designing future meningococcal vaccines. Such an approach has the advantages that it uses data derived from human, rather than animal, immunization and that it avoids the need to screen individual antigens.