Pre-clinical evaluation of pneumococcal vaccine candidate molecules in a mouse pneumococcal pneumonia model

Abstract

Background: Streptococcus pneumoniae (pneumococcus) is a major cause of morbidity and mortality in young children. Pneumococcal conjugate vaccines are restricted in their serotype coverage, and hence, focus has been directed towards common protein-based antigens. However, there have been problems associated with testing protein-based antigens in animal models. These include inhumane endpoints, use of large sample sizes, and the inability to detect subtle differences between antigens. This study describes a novel intranasal pneumonia co-inoculation model conducted in mouse and aims to address some of these limitations. Methods: Eighty female NMRI mice were equally distributed into the vaccinated and placebo groups. The mice were subcutaneously inoculated with recombinant PspC or PpmA or adjuvant (control). Both groups were co-inoculated with 50 μL of a bacterial suspension containing D39:: rpsL (serotype 2) or PJ351:: rpsL (serotype 1) wild-type strains and their isogenic knockout strains containing a trimethoprim resistance marker. Mice (n = 5 per group) were terminated at 0, 6, 12 and 24 hours post-inoculation. Bacterial samples were obtained from nasopharyngeal washings, bronchoalveolar lavage fluid (BAL fluid) and homogenised lung tissue. Bacterial counts were obtained from selective media containing either streptomycin or trimethoprim, and the ratios of wild-type and knockout were determined. Serum was collected by cardiac puncture and ELISA was used to monitor the levels of IgG antibodies. To further refine the model, a single time point (T = 24h) in which events were more pronounced was chosen for the rest of the experiments and the number of mice was scaled down to 20 mice per experimetal procedure. A group of 20 mice (10 in placebo and 10 in vaccinated mice) were subcutaneously vaccinated with either recombinant PspA, IgA protease and SlrA or adjuvant (control). Both vaccinated and adjuvant groups were co-inoculated with 50 μL of a bacterial suspension containing D39:: rpsL (serotype 2) or PJ351:: rpsL (serotype 1) wild-type strains and their isogenic knockout strains. Bacterial samples were collected from mice as described above. Results: Pneumococcal knockouts of D39 and PJ351 lacking the genes encoding PpmA and PspC were attenuated in their ability to colonise the nasopharynx, BAL fluid and lungs of mice. Knockouts deficient in PspA, IgA bacterial protease and SlrA in the PJ351 background were attenuated in all sites sampled, whereas knockouts in the D39 background were attenuated only in the nasopharynx. PspC IgG antibodies were able to elicit clearance of PspC producing wild-type D39 and PJ351 strains more effectively compared to PspC deletion knockouts, which were able to persist in the nasopharynx, BAL fluid and lungs. Responses in PpmA vaccinated mice were dependent on the genetic background of the inoculation strains. Clearance of D39 wild-type strains was more noticeable in the nasopharynx than in the BAL fluid and lungs, whereas the wild-type PJ351 was cleared more efficiently than the knockout in the BAL fluid and lungs, compared to the nasopharynx. Responses in IgA bacterial protease vaccinated mice were also dependent on the genetic background of the inoculation strains, with a clearance of the D39 wild-type strain being more noticeable in the nasopharynx. PspA and SlrA IgG antibodies were not protective against infection with both strains of the pneumococcus in the mouse model. Conclusion: Using PspC, PpmA, PspA, IgA bacterial protease and SlrA as examples, we demonstrated that this mouse co-colonisation model could effectively detect subtle differences in the clearance of wild-type and knockout bacteria in the presence of antibodies against non-essential surface antigens. The model is humanely more acceptable, as it uses far fewer animals than conventional testing, with greater sensitivity. Further, it relies on measurements of bacterial density, rather than fatal disease as endpoints. The model also provides additional benefits in terms of monitoring the virulence characteristics of the target antigens.