Construction of a genome scale metabolic model of Mycobacterium tuberculosis to investigate growth-regulated modulation of metabolism.

The TB bacillus is an important pathogen of man and animals that kills about three million people each year. This proposal is to develop a virtual model of the BCG vaccine strain of the TB bacillus in a computer. The model will provide important insight into how this pathogen grows and replicates. The model may also be used to perform virtual experiments that would be very hard or impossible to perform in the real world. For instance, it is difficult to study the behaviour of the TB bacillus when it is growing inside patient's lungs. But if we can identify the metabolic pathways that are active then we can perform easily virtual experiments that will, for instance, investigate how the bacillus will respond to a new antibiotic whilst it is living in the patient's lungs. The virtual TB may also be used to screen new compounds for activity against the TB bacillus: the effect of various antibiotics can be tested in the virtual TB cell far more quickly than with live cells (and with no possibility of the experimenter catching TB). But probably most importantly, the virtual TB cell can be used to invent new antibiotics, by identifying pathways, or groups of pathways, that are essential for growth. This model will first be built using DNA sequence data from the genome. However, to make the virtual cell more realistic, we must incorporate biological data. We will therefore grow the real life organism (actually the vaccine strain of the TB bacillus) in highly defined conditions in the laboratory and perform chemical analysis of what goes in and what comes out of the cell. A remarkable mathematical technique, known as metabolic flux analysis, can then be used to estimate the flux of metabolites through each central metabolism pathway inside the cell. This information will be incorporated into the virtual cell to make its behaviour correspond more closely with the biological organism. The next stage of the project is testing our virtual cell. To do this we will identify which pathways are essential in the virtual cell and then inactivate those pathways in living cells. We will then see if the growth (or absence of growth) of the real life cells matches the predictions of the model. These tests will be used to refine and improve the model that may thereafter be used in drug development.

The TB bacillus is an important pathogen of man and animals. New drugs are badly needed, particularly drugs that are effective against the persistent slow-growing form of the bacillus. However, very little is known concerning the physiological state of the tubercle bacillus during persistence. We have initiated a project to investigate Mycobacterium tuberculosis metabolomics for fast and slow growing cells and demonstrated differences in metabolism that may be relevant to persistence. This investigation is to further these findings and investigate the hypothesis that the tubercle bacillus modulates its metabolism in response to changes in growth rate. A genome-scale in silico M. tuberculosis model will initially be constructed using the genome annotation. Although useful, the model will be inadequate for in silico simulations, as information on the relative flux metabolites through each pathway cannot be extracted from genome data; and there are many orphan genes and genes with uncertain homology to genes of known function. We will therefore constrain the in silico model with (i) extracellular metabolite and macromolecular data to construct a central metabolism metabolic flux model; (ii) intracellular metabolite data to construct a more complete central metabolism flux model including parallel, divergent and bidirectional intracellular pathways. Metabolite levels will be measured at the MeTRO laboratory by a combination of biochemical and spectroscopy techniques including GC-MS, NMR-MS, LC-NMR and LC-MS. For intracellular metabolite measurements we will perform carbon-labelling experiments using 13C-labelled substrates. The model will be further interrogated by (i) targeted gene deletion and further flux measurements; (iii) whole genome transposon mutagenesis data indicating relative fitness of a library of mutants at both growth rates. The data will be integrated into the model to build a realistic full-scale metabolome/genome model of the TB bacillus at fast and slow growth rates. The model will be used to test the target hypothesis and may be further developed for drug development and analysis of transcriptome and metabolome data.