**2. A historical perspective of studies on** *M. tuberculosis* **persistence**

Early glimpses of the unique adaptability of *M. tuberculosis* appeared in two landmark studies conducted in the early 20th Century. First, Corper and Cohn observed that 24 out of 56 *in vitro* cultures of human and bovine isolates contained culturable tubercle bacilli even after 12 years of incubation in sealed containers (Corper & Cohn, 1933). This *in vitro* study revealed the characteristic persistence of *M. tuberculosis* in bacteriostatic condition. Concurrent with this *in vitro* study, Opie and Aronson reported the presence of virulent *M. tuberculosis* bacilli in about 26% of lesions resected from individuals dying of causes unrelated to TB (Opie & Aronson, 1927). While this study demonstrated asymptomatic infection of *M. tuberculosis*, it also opened up questions as to how bacilli are able to evade the immune system and suppress inflammation. In subsequent follow-up studies it appeared that the bacilli were unexpectedly present in uninvolved tissues instead of the presumed primary lesions (Feldman & Baggenstoss, 1939). These *in vivo* studies thus raised speculations that a competent immune system is capable of clearing the bacilli at the primary lesions, but the bacilli could have used escape mechanism to survive at secondary sites in presumably a non-replicating state.

The persistent nature of *M. tuberculosis* re-occupied the spotlight of tuberculosis research during the early phase of antibiotic-era around mid-20th Century. In several bacteriological studies on resected lesions from antibiotic treated individuals the bacilli could be microscopically observed even though individuals had converted to sputum negative (Loring & Vandiviere, 1956, Loring *et al.*, 1955, Vandiviere *et al.*, 1956, 1953). Interestingly, these bacilli in many instances were non-culturable and often associated with resolved lesions, thus raising a debate whether these were dead, or viable but non-culturable bacilli. The idea of viable but non-culturable bacilli seemed more convincing after McDermott and colleagues demonstrated reactivation of TB in mice upon termination of chemotherapy that was sufficient to reduce viability to undetectable levels (McCune *et al.*, 1966). It was, however, not clear in this study as to how and where the viable bacilli persisted, but the correlation between non-culturable bacilli and closed hypoxic lesions (Vandiviere et al., 1956, Haapanen *et al.*, 1959) support the idea that closed lesions could possibly be the primary site of non-replicating persisters which developed in the bacteriostatic environment of the lesions.

Several attempts have been made to investigate the physiology of non-replicating persisters using *in vitro* models such as hypoxic and nutritionally starved cultures. These studies subsequently led to identification of genetic components responsive to these conditionsmost notably *isocitrate lyase* (*icl*) of the glyoxalate shunt pathway and the two-component regulatory system *dosR-dosS (Park et al., 2003, Hobby & Lenert, 1957, Wayne & Hayes, 1996, Wayne & Lin, 1982, Saini et al., 2004)*. While the mutation in *icl* impairs the persistence of bacilli in a mice model (McKinney *et al.*, 2000), the phenotype of *dosR-dosS* mutants in animal models have yielded conflicting results (Rustad *et al.*, 2008, Parish *et al.*, 2003, Malhotra *et al.*, 2004).
