How Mycobacterium tuberculosis goes to sleep: the dormancy survival regulator DosR a decade later

Some bacteria form spores when under stress. In Bacillus subtilis, the response regulator Spo0A plays a key role in this developmental process [1]. Mycobacteria, including the notorious tubercle bacillus, are not capable of forming spores [2]. However, they have a genetic program, triggered by hypoxia and respiratory poisons and controlled by the transcription factor DosR [3], which regulates the development of a defined nongrowing survival form without morphological differentiation. This quiescent physiological state maintains viability for extended periods of time and, importantly, shows phenotypic (as opposed to genetic) drug resistance. Thus, it is thought that this stage of the bacillus life cycle contributes to both key issues we face in the control of TB, the symptom-free latent state of TB infection and the persistence of active disease despite prolonged (6 months to years) chemotherapy [4–6]. Because of the possible importance of dormant (i.e., nongrowing) bacilli in pathophysiology and chemotherapy, as well as for the development of vaccines, biomarkers and diagnostics for TB, the anaerobic nonreplicative form of the obligate aerobe and its genetic dormancy program have drawn considerable interest since the discovery of DosR a decade ago [3,6,7]. Antidormancy drugs might not only be the key for eradication of latent infection, but also for the shortening of the treatment time for the active disease [8]. In the following review, we discuss what we have learnt since the discovery of DosR, the gaps in our knowledge and a possible relevance of DosR in emerging mycobacterial pathogens.

Discovery of DosR

DosR was discovered in a classical reverse genetic approach. A proteomic analysis of the hypoxia-induced dormancy response in the BCG strain of the tubercle bacillus identified a 23-kDa candidate response regulator (Rv3133c), together with three other proteins (α-crystallin-like protein Rv2031, USP domain-containing Rv2623 and CBS domain-containing Rv2626c), to be upregulated immediately upon entry into dormancy. The upregulation of these dormancy proteins coincided with the induction of their transcription and was not observed in the standard aerated stationary phase, thereby suggesting specificity and a functional role for this transcription factor in the hypoxia-induced dormancy response [9,10]. Genetic analysis demonstrated that the loss of function of the response regulator resulted in a dramatic loss of viability of nongrowing hypoxic-dormant cultures, as well as the loss of induction of the coinduced dormancy proteins. Based on this dual phenotype – loss of viability and regulation – the response regulator was named ‘dormancy survival regulator’, or DosR [3]. The genetic program underlying the dormancy response had been identified. The loss of viability of the dosR mutant clearly demonstrated that DosR and its regulon are key for adaptation of the bacillus to nonreplicating survival in hypoxic environments suspected (and later directly demonstrated [11,12]) in TB lesions in vivo.

It is useful to note that there is some confusion in the literature and databases regarding the nomenclature of dosR, which is also sometimes called devR. dosR was first identified in 1998 in the Mycobacterium tuberculosis H37Rv genome project by Cole and colleagues, where it was named Rv3133c and annotated as a putative response regulator based on bioinformatic analyses [13]. In 2000, Dasgupta et al. searched for genes that are differentially expressed in the avirulent H37Ra strain versus the virulent H37Rv strain of the tubercle bacillus [14]. Rv3133c appeared to be differentially expressed, and the authors named the gene devR for ‘differentially expressed in the virulent strain’. However, Dasgupta and colleagues then showed that only two out of three HRv37a isolates displayed a reduced expression level. A third isolate of the attenuated strain showed a level of expression of both mRNA and protein that was identical to that of H37Rv [14]. Finally, in 2002, after the discovery that Rv3133c is induced in the tubercle bacillus by hypoxia and assigning physiological and regulatory functions to the gene based on genetic analyses as described above, we named the gene dosR[3,10].

DosRST two-component system & the DosR regulon

Transcriptomic analyses showed that DosR controls a regulon of approximately 50 genes: the dormancy or DosR regulon [15–17]. Further work revealed that the regulon is activated – in addition to oxygen starvation – by two signals produced inside host lesions: nitric oxide [18] and carbon monoxide [19,20]. The sensors controlling the DosR transcription factor are the histidine kinases DosS and DosT, representing redox and hypoxia sensors, respectively. In addition, the two kinases function as detectors of the respiration-impairing gases nitric oxide and carbon monoxide [3,18–24]. More recent work linked DosR and its regulon to additional signaling networks involving Ser/Thr protein kinases [25]. The precise functions of all the members of the DosR (protein-coding) regulon still remain to be determined (see [7,26–28]). This is also the case for the recently discovered noncoding small RNAs that appear to be under DosR control [29]. Physiological analyses of a M. tuberculosis dosR-knockout strain demonstrated that the massive loss of viability correlates with a collapse of energy and redox homeostasis, thus suggesting that the DosR regulon encodes critical functions required for adaptation to the reduced metabolic state, which is characteristic of nonreplicating tubercle bacilli [30]. It was also demonstrated that DosR is required for resumption of growth once M. tuberculosis exits the nonreplicating state. This suggests that the dormancy regulon is not only essential for dormancy survival per se, but also for successful transition between (oxygen) respiring and nonrespiring conditions without loss of viability [30]. The importance of maintaining energy and redox homeostasis during dormancy, and the critical function of anaerobic respiration for the latter, was shown via genetic and chemical probing [31,32]. It is notable that the new clinical candidate TMC207, a diarylquinoline ATP synthase inhibitor that kills dormant bacilli, showed very good efficacy in multidrug-resistant TB patients in a proof-of-concept study [33,34]. The strong efficacy of TMC207 is attributed, at least in part, to the role of its ATP synthase target in maintaining the critically low levels of ATP, which is the Achilles heel of nonreplicating bacilli [31,35]. Metabolic analyses of anaerobic-dormant bacilli revealed major rearrangements of central metabolism, including the utilization of the reductive branch of the tricarboxylic acid cycle. The adoption of intracellularly generated fumarate as an electron sink with subsequent secretion of succinate suggests that the obligate aerobe M. tuberculosis employs fermentative processes for maintaining an energized membrane and viability when oxygen as an external electron acceptor becomes limiting [36]. Regarding the functional role of the original members of the DosR regulon [10], it is interesting to note that the ATP-binding, USP domain-containing dormancy protein Rv2623 has the ability to regulate growth in vitro and in vivo, and is required for the establishment of a persistent infection [37]. The α-crystalline-like gene appears to have somewhat similar growth-related functions [38]. The CBS domain-containing Rv2626c protein was found to modulate macrophage effector functions by eliciting both innate and adaptive immune responses, suggesting its possible use as a vaccine candidate [39]. Its role in bacterial physiology remains to be determined.

Conservation, expression & function of the DosR regulon in vivo

DosR, although genetically not essential for growth in vitro, is conserved throughout clinical isolates of the tubercle bacillus, pointing to its importance for in vivo biology of the obligate pathogen. The relevance of the DosR regulon for infection and disease was further demonstrated by its induction in macrophages (the tubercle bacillus is a facultative intracellular parasite) [15] and in several animal models for TB infection such as mice and guinea pigs [18,40]. Importantly, it was shown in humans that the DosR regulon proteins are expressed in both latent infection and active TB disease [41–44]. Further evidence for clinical relevance comes from a study showing that M. tuberculosis in sputum expresses the DosR regulon [45].Surprisingly, studies of DosR mutants in animal models showed various phenotypes ranging from hyper- to hypo-virulent forms [46–50]. The reasons for these variable outcomes are not entirely clear but might be due to the different lesion types encountered in the various animal models (e.g., mice do not form human-like hypoxic granulomas), the various bacterial and animal strains used and differences in the genetic constructs and routes of infection [49]. It is important to note that analyses of DosR function in rabbit and guinea pig models of TB (both displaying a more human-like histopathology and hypoxic lesions) showed attenuation, as one might expect based on the in vitro phenotypes of the dosR loss-of-function mutants [49,50]. These results point to the requirement of DosR for persistence of M. tuberculosis in humans. However, a decade after the initial discovery, the function of DosR in human latency and disease is still not fully understood, reflecting that M. tuberculosis is a very slow-growing parasite, and that we still struggle with predictive and practical animal models for TB research. To finally determine the relevance of DosR in TB, it might be necessary to carry out the analyses in nonhuman primates [51] (i.e., in a predictive animal model that reproduces the spectrum of active disease and latency we see in humans and shows the spectrum of lesion types observed in humans). These studies need to address the effect of loss of dosR function on the efficacy of standard chemotherapy of active disease and latent infection. The prediction is that loss of dosR function will result in a persistence phenotype in latent infection and active disease (perhaps including transient hypervirulence) and in a more rapid elimination of bacilli during prophylactic and therapeutic chemotherapy, bearing in mind that the effects could be lesion type-dependent. It is interesting to note that evidence for a clinical and therapeutic relevance of hypoxic nonreplicating bacilli comes from in vivo probing with metronidazole, a drug that shows strict (albeit weak) selective toxicity for the hypoxic-dormant form of the bacillus in vitro. Aerobically growing bacilli are not affected by the drug [5]. Studies in rabbits – a TB animal model presenting hypoxic lesions – showed that metronidazole significantly reduces the number of bacilli in the lung [11]. These results are consistent with metronidazole studies in patients showing that this dormancy-specific drug has some effects on the disease [52]. An ongoing metronidazole Phase II clinical trial in Korea, sponsored by the National Institute of Allergy and Infectious Diseases (NIAID) and using high-resolution computed tomography imaging as an end point of the responses of specific lesions to chemotherapy, might shed further light on the clinical relevance of hypoxic nonreplicating bacilli and the DosR regulon [101].

Future perspective

While struggling to determine the details of the dosR signaling pathway, functions of the regulon and to understand as well as confirm the connection between clinical persistence/latency and bacterial dormancy in the global health emergency that is TB, a new mycobacterial public health threat is raising its head: pulmonary diseases caused by nontuberculous mycobacteria (NTM) [53]. Although the epidemiology is not well understood, the incidence of diseases caused by these ubiquitous soil and water mycobacteria appears to be increasing. Municipal water systems appear to be one of the sources of infection [53,54]. Interestingly, in the treatment of NTM disease, we see the same clinical phenomenon that we observe for TB: potency in standard in vitro antimicrobial susceptibility assays does not predict rapid clinical response. Treatments can go on for years to reach cure, if cure is ever achieved. For several of the NTM diseases, such as (drug-susceptible) Mycobacterium abscessus lung infections, surgery is recommended if possible [55–57]. The in vitro–in vivo potency disconnect of drugs appears to be a hallmark of all mycobacterial infections. Comparative genomic analysis has demonstrated the conservation of DosR and the dormancy regulon across mycobacterial species, including NTM [58]. Interestingly, expression of NTM dormancy proteins was demonstrated in human NTM infection [59]. Genomic conservation, together with in vivo expression of the NTM DosR regulon, suggest that the clinical persistence of NTM disease, despite extensive chemotherapy, could be caused – similarly to TB – by bacterial dormancy. Whether NTM can undergo a dormancy response in vitro has not been demonstrated. However, for the fast-growing saprophytic soil organism Mycobacterium smegmatis, it was shown that this species is capable of undergoing a hypoxia-induced dormancy response with striking similarities to that observed for M. tuberculosis. Upon encountering hypoxia, M. smegmatis exits the cell cycle and develops into a long-lasting, phenotypically drug-resistant, metronidazole-sensitive, diploid survival state that resumes synchronized cell division upon reintroduction of oxygen [60–62]. This suggests that the dormancy response is an evolutionarily old response in mycobacteria that possibly developed to deal with fluctuating oxygen concentrations in the soil. It is tempting to speculate that once mycobacteria infect animals and humans, they employ their soil-evolved dormancy program to deal with hostile conditions (e.g., nitric oxide and carbon monoxide) and fluctuating oxygen availability (hypoxic lesion microenvironments) in their hosts. The fact that sleeping hypoxic mycobacteria also coincidentally evade our growth inhibition-directed – and, importantly, oxygen-dependent [63] – antibacterials is an unfortunate side effect.