26 October 2007
Caroline Colijn
Department of Epidemiology
Harvard School of Public Health
Tuberculosis (TB) has become a global health challenge, causing millions of deaths per year worldwide. It is caused by infection with the bacterium Mycobacterium tuberculosis, and is transmitted through the respiratory route. While it is estimated that 1/3 of the world's population is infected with TB, most people will contain the infection in a latent (asymptomatic and uninfectious) state indefinitely. Those who do develop active disease may do so soon after infection, or many years later, upon which their treatment requires multiple antibiotics given over at least a 6 month period. In recent years, multi drug resistant and extensively drug resistant (MDR and XDR) TB strains have emerged which are resistant to isoniazid and other therapeutic agents. Drug resistance threatens to undermine global TB control efforts, particularly in locations with high HIV prevalence, so that the development of new therapies that could provide effective treatment with a shorter regimen, and therapies effective against MDR and XDR strains, is a high priority in TB control. While our knowledge of the mechanisms of action even of current therapeutic agents remains incomplete, several first-line anti-TB drugs including isoniazid target the synthesis of mycolic acids, fatty acids essential for survival and proliferation of M. tb which form part of the mycobacterial cell wall.
The discovery of new drug targets will depend critically on understanding the metabolic response to the disruption of target function. This response depends on both gene regulatory effects influencing expression, and on the response of intracellular metabolic networks to the induced variation in protein levels. We present an approach to integrating gene expression data with in silico metabolic models, which we hope will assist in linking gene expression data to phenotypic behavior and in predicting the effects of disrupting targets in metabolic pathways. We apply the method to the mycolic acid synthesis pathway in M. tuberculosis, and are able to predict the capability of M tb to produce mycolic acids under a wide variety of in vitro conditions based on published expression data. Predictions for isoniazid and ethionamide, as well as several other drugs and compounds, agree with the measured affects of these agents on mycolic acid biosynthesis, providing a proof of principle that the approach enables the prediction of metabolic effects from gene expression data.