The bacterium that causes tuberculosis is one of the most successful pathogens. Its spread among humans has been so efficient that as much as one-third of the world's population is now believed to be infected1. In most cases, these infections cause clinically silent disease, which is likely to remain permanently dormant unless the host's immunity is substantially compromised. The extraordinary stealth and opportunism that Mycobacterium tuberculosis exhibits results from the complex and delicately coordinated way in which it interacts with its host — a process controlled in part by a specialized bacterial protein-secretion system called ESX-1. On page 717 of this issue, Raghavan et al.2 provide insight into how mycobacteria might maintain ESX-1 activity at just the right level, perhaps allowing these pathogens to avoid excessive virulence so that they can patiently pursue their low-key yet incredibly effective strategy for long-term survival.

Mycobacteria lack the specialized type I–VI secretion systems that have been well characterized in Gram-negative bacteria. Instead, their virulence is mediated in part by ESX-1, which is also found in other Gram-positive bacteria. Two proteins secreted by the ESX-1 system, ESAT-6 and CFP-10, are prominent targets of the immune system in animals, including humans, infected with M. tuberculosis3. These proteins are encoded together as part of a coordinated unit of genetic material, the esxBA operon. Genomic studies4 predicted that genes surrounding esxBA are likely to encode components of a secretory apparatus responsible for the export of ESAT-6 and CFP-10 out of the bacterial cytoplasm. This prediction was soon validated5,6, defining ESX-1 as a prototypical virulence-associated mycobacterial protein-secretion system.

Exactly how ESX-1 contributes to mycobacterial virulence remains a puzzle, although various observations have led to several hypotheses that are not mutually exclusive. For example, ESX-1 mutants grow poorly during the early phase of infection in mice and in macrophages6,7 (immune cells that engulf pathogens by phagocytosis). Moreover, in infected dendritic cells (another type of phagocytic immune cell), ESX-1 helps M. tuberculosis escape from phagosome vesicles — in which they are captured for destruction — into the cytoplasm, where they can replicate8. Because purified ESAT-6 acts as a membrane-disrupting toxin in vitro9, this protein could directly mediate phagosomal escape. However, ESX-1 also hinders phagosome maturation in macrophages through a process that does not require ESAT-6 or other known substrates secreted by this system10. Taken together, these data indicate that ESX-1 modulates innate immune responses of the infected host through several mechanisms, which probably involve ESX-1-mediated secretion of ESAT-6 and other unidentified factors.

Raghavan and colleagues2 identify a previously unknown component of ESX-1 that is not only a central regulator of it but is also secreted by this system. They find that, in M. tuberculosis, the Rv3849 gene — which is located some distance from the main ESX-1 gene cluster surrounding the esxBA operon — is required for ESX-1 function. The protein product of Rv3849, EspR, is highly similar to a gene transcription factor of the harmless soil bacterium Bacillus subtilis. The authors find that EspR is also a DNA-binding transcriptional regulator — Rv3849 deactivation leads to changes in the transcription of a few operons in the M. tuberculosis genome, including the Rv3616c–3612c cluster, which encodes at least two ESX-1-secreted proteins and is required for the functioning of ESX-111 (Fig. 1).

Figure 1: Genes involved in ESX-1 secretion.
figure 1

At least three separate regions of the M. tuberculosis chromosome encode genes that are involved in the ESX-1-secretion system. The esxBA operon encodes the main secreted products ESAT-6 and CFP-10. Raghavan et al.2 identify the protein product of the Rv3849 gene — which is some distance (19.4 kilobases) away from the esxBA operon — as the EspR protein. This protein is both a secreted substrate of ESX-1 and a positive transcriptional regulator of a third region, the Rv3616c–Rv3612c operon, which is also involved in ESX-1-mediated virulence. Genes encoding known secreted substrates of ESX-1 are shown in green, and those encoding structural or functional components of the ESX-1 secretion apparatus are in red. Genes in white are of unknown function, although they are presumed to be somehow involved in the ESX-1 system.

Although it is not surprising that mycobacteria express a transcription factor that regulates the expression of ESX-1 components, that EspR is itself secreted by ESX-1 is an unexpected result. The authors propose that secretion of this protein constitutes an unusual feedback loop that could be part of a finely tuned control process to prevent excessive or prolonged activity of ESX-1 during infection of mammalian cells. Although secretion of a regulatory protein as a mechanism for diminishing its activity inside the cell has been described before, EspR might represent the first example of a transcription factor that is actively exported from the cell by the same secretion system that it induces.

One can imagine how EspR could impose a limit on the level and duration of ESX-1 activity. When initially expressed, this protein could be essential for activating ESX-1 to secrete high levels of virulence-promoting proteins, thus allowing M. tuberculosis to establish its infection in the host. Once infection is achieved, secretion of EspR might lead to a reduction of its transcriptional activity within the bacterium and diminished ESX-1 activity. This would partially attenuate virulence, thus favouring either bacterial persistence or a slow, chronic infection in order to enhance transmission.

Can EspR production be turned on or off, and — if so — what external stimuli and bacterial sensing and signalling molecules could be responsible for this? Are other components of the ESX-1 system separately regulated by factors distinct from EspR? These questions, together with the identification of other secreted ESX-1 substrates and their mechanisms of action in mammalian host cells, should provide many opportunities for deciphering the unique logic of mycobacterial virulence strategies.

It will also be interesting to determine whether EspR has any specific function once it has been exported from the bacterial cell. Because at least two other ESX-1 secreted products — ESAT-6 and CFP-10 — are major targets for the immune response, it is possible that EspR is also a prominent mycobacterial antigen. If so, this could have implications for the development of new vaccines and diagnostic tools for tuberculosis.