We have studied two very different animal models of viral infection: a primate model of neuroAIDS, and a mouse model of viral immunology. In the neuroAIDS model, rhesus macaques are infected with SIV, eventually develop encephalitis and die. Cerebrospinal fluid is used for metabolomics because it reflects the biochemical state of the brain, and is an accessible biolfluid in humans. The second system is LCMV infection in mouse, which produces a rapid immune response in a well-defined time frame.

A global metabolomics approach was used to study the neurochemical effect of SIV infection in rhesus macaques, a model system for HIV and neuroAIDS, and illustrates the potential of metabolomics to address problems in central nervous system biochemistry and neurovirology, as well as neurodegenerative diseases. Cerebrospinal fluid (CSF) was compared before and after viral infection, and more than 3,500 features were measured. There were significant changes in the metabolome, with a general increase in metabolite concentrations during infection. Specific metabolites which changed were identified using database searching and comparison of the MS/MS pattern using a QTOF. Fatty acids, including palmitic, myristic, oleic, and stearic acids increased during infection, as did the corresponding lysophosphocholines. Other molecules that increased significantly during infection included carnitine, and the acylcarnitines, octanoylcarnitine and butyrylcarnitine. All of these molecules are related to fatty acid and lipid metabolism.

Gene chip experiments were then performed on the hippocampus of uninfected compared to infected animals. These results indicated that different phospholipase genes, including PLA1 and PLA2 were up-regulated during infection. This increase in transcript levels was confirmed using quantitative PCR.

Finally, specific biochemical assays were performed on brain tissue, indicating that phospholipase 2 activity was indeed increased in the brain as a result of viral encephalitis. Thus, the metabolomics results were able to generate a testable hypothesis, which was confirmed by using gene chip experiments and biochemistry. These three datasets reinforce each other, and demonstrate the process of going from descriptive metabolomic phenomenology to a testable biochemical hypothesis, even in a disease as complex as neuroAIDS.

The combination of metabolomics, transcriptomics and biochemistry, greatly facilitates the process of understanding metabolomic changes in a biological context. Other specific approaches can be used, such as immunohistochemistry, to localize changes at tissue and cellular resolution, or to determine substrate binding and kinetics in and ex vivo model, as was done with the kidney transporters. The use of metabolomics to understand and actually solve complex biomedical and biochemical problems using animal models will be emphasized. The integration of various techniques with metabolomics to understand these complex systems will be emphasized.