image_3.pdf (351.06 kB)


Download (351.06 kB)
posted on 23.03.2018 by Stéphanie Ghosn, Soulaima Chamat, Eric Prieur, Antoine Stephan, Pierre Druilhe, Hasnaa Bouharoun-Tayoun

The lack of preclinical models able to faithfully predict the immune responses which are later obtained in the clinic is a major hurdle for vaccines development as it increases markedly the delays and the costs required to perform clinical studies. We developed and evaluated the relevance to human immune responses of a novel humanized mouse model, humanized-spleen cells-NOD-SCID-gamma null (Hu-SPL-NSG), in which we grafted human spleen cells in immunodeficient NOD-SCID-IL-2rγnull (NSG) mice. We selected the malaria vaccine candidate, Liver Stage Antigen 3-Full Length, because we had previously observed a major discrepancy between preclinical and clinical results, and compared its immunogenicity with that of a shorter form of the molecule, LSA3-729. NSG mice engrafted with human spleen lymphocytes were immunized with either LSA3-FL or LSA3-729, both adjuvanted with montanide ISA720. We found that the shorter LSA3-729 triggered the production of human antibodies and a T-helper-type 1 cellular immune response associated with protection whereas LSA3-FL did not. Results were consistent in five groups receiving lymphocytes from five distinct human donors. We identified antigenic regions in the full-length molecule, but not in the shorter version, which induced T-regulatory type of cellular responses. These regions had failed to be predicted by previous preclinical experiments in a wide range of animal models, including primates. Results were reproducible using spleen cells from all five human donors. The findings in the Hu-SPL-NSG model were similar to the results obtained using LSA3-FL in the clinic and hence could have been used to predict them. The model does not present graft versus host reaction, low survival of engrafted B lymphocytes and difficulty to raise primary immune responses, all limitations previously reported in humanized immune-compromised mice. Results also point to the shorter construct, LSA3-729 as a more efficient vaccine candidate. In summary, our findings indicate that the Hu-SPL-NSG model could be a relevant and cost-saving choice for early selection of vaccine candidates before clinical development, and deserves being further evaluated.