A Bayesian Change point model for differential gene expression patterns of the DosR regulon of Mycobacterium tuberculosis
1 School of Crystallography, Birkbeck College, University of London, Malet Street, London, WC1E 7HX, UK
2 TB research, Health Protection Agency, CEPR, Porton Down, Salisbury, SP4 0JG, UK
3 MRC Biostatistics Unit, University Forvie Site, Robinson Way, Cambridge, CB2 0SR, UK
BMC Genomics 2008, 9:87 doi:10.1186/1471-2164-9-87Published: 22 February 2008
Low oxygen availability has been shown previously to stimulate M. tuberculosis to establish non-replicative persistence in vitro. The two component sensor/regulator dosRS is a major mediator in the transcriptional response of M. tuberculosis to hypoxia and controls a regulon of approximately 50 genes that are induced under this condition.
The aim of this study was to determine whether the induction of the entire DosR regulon is triggered as a synchronous event or if induction can unfold as a cascade of events as the differential expression of subsets of genes is stimulated by different oxygen availabilities.
A novel aspect of our work is the use of chemostat cultures of M. tuberculosis which allowed us to control environmental conditions very tightly. We exposed M. tuberculosis to a sudden drop in oxygen availability in chemostat culture and studied the transcriptional response of the organism during the transition from a high oxygen level (10% dissolved oxygen tension or DOT) to a low oxygen level (0.2% DOT) using DNA microarrays. We developed a Bayesian change point analysis method that enabled us to detect subtle shifts in the timing of gene induction. It results in probabilities of a change in gene expression at certain time points. A computational analysis of potential binding sites upstream of the DosR-controlled genes shows how the transcriptional responses of these genes are influenced by the affinity of these binding sites to DosR. Our study also indicates that a subgroup of DosR-controlled genes is regulated indirectly.
The majority of the dosR-dependent genes were up-regulated at 0.2% DOT, which confirms previous findings that these genes are triggered by hypoxic environments. However, our change point analysis also highlights genes which were up-regulated earlier at levels of about 8% DOT indicating that they respond to small fluctuations in oxygen availability. Our analysis shows that there are pairs of divergent genes where one gene in the pair is up-regulated before the other, presumably for a flexible response to a constantly changing environment in the host.