Global Mapping of Molecular Networks Revealed by Integrative Omics Analysis Highlights Potential Mechanisms in a Non-Human Primate Model of Malaria

Background: Host response to malarial infection has not been well characterized systematically on “multi-omics” levels. Unveiling mechanisms across molecular layers is crucial for understanding pathology in affected tissues. Materials and Methods: We report an integrative analysis of transcriptomics, metabolomics, and lipidomics in peripheral blood (PB) and in bone marrow (BM) using a Non-Human Primate (NHP) animal model infected with Plasmodium cynomolgi (a model for Plasmodium vivax infection in humans)over 100 days. We developed a framework to integrate longitudinal “multi-omics” data types that account for potential nonlinear relationships among the heterogeneous data types using a mutual information-based machine learning approach. Inferred “multi-omics”networks were further correlated to immunological profiles and clinical traits using biweight midcorrelation to bridge “multi-omics” molecular layers with phenotypic layers.Results An atlas of tissue-specific network modules of malaria was generated, which revealed different mechanisms of dysregulation on arms of immune system in PB and BM. Integrated “omics”analysis showed temporal and tissue-specific dynamics of the host response to P. cynomolgi infection. Profiles of lipidomics and metabolomics correlated with parasitemia and clinical traits implies roles of leukotriene and lipoxin in inflammatory regulation at acute primary infection. Correlation analysis of modules inferred from integrative analysis of the “omics” datasets with immunological profiles and clinical traits implicates the impact of PD-1 signaling on a subset of CD8+ T cells, and suggests the involvement of mTOR pathway in host response in a distinct immune cell population. While mTOR involvement in host response has been shown in a mouse model of malaria, to date it has not been shown in NHP or human malaria studies. The results suggest that PD-1 and mTOR might be novel therapeutic targets for malaria treatment. Conclusions: To our knowledge, this is the first report of large-scale, longitudinal “multi-omics” experimental datasets generated using a NHP malaria model. This global “multi-omics” map of malaria pathology provides a unique resource for malaria study, and illustrates the power of integrative “multi-omics” analysis to characterize key processes in malaria.