**5. Phenotypic Drug-Resistance of Mtb**

Caseous granulomas and the cavities of the lungs of TB patients harbor subpopulations of NR bacilli which are phenotypically drug-resistant but genetically susceptible, commonly referred to as persisters. Characterized by a transient, non-heritable drug tolerance, persisters are capable of withstanding bactericidal drug concentrations, and once the antibiotic is removed, to resume growth with genetic features identical to the original strain.

The level of resistance to different antimicrobial agents varies with the in vitro stress model used [31–34,74–77], including hypoxia (Wayne dormancy model) [78,79], nutrient starvation [80], acids and/or nitric oxide [81,82], stationary phase [83], antibiotic-starved strains [84] and others, or their variants. In the Wayne model, in which dormant bacilli are obtained by a gradual adaptation to anaerobiosis through the self-generated formation of an oxygen gradient, nonreplicating persistence (NRP) stages 1 and 2 were observed [78]. NRP-2 cells developed a thickened outer layer that helped in restricting RIF entry [85]. Our group used the Wayne model at different pHs: pH 6.6, the pH or culture media [86], pH 5.8, to mimic the environment of cellular granulomas [87], pH 7.3, to mimic the environment of caseous granulomas [88]. We found that at pH 5.8, several drugs killed NR bacilli, with the best being the rifamycins RIF and rifapentine (RFP), while at pH 7.3, only RIF and RFP killed dormant bacilli out of 12 drugs tested [88]. Since the rifamycins were the only agents sterilizing caseum obtained from rabbits [35,36], our model could mimic caseum to measure drug activity against NR Mtb in this environment. In hypoxia at pH 7.3, we found that RIF plus nitazoxanide (a nitro-compound for anaerobic infections) killed NR Mtb cells, while the combination currently used for human TB therapy (RIF-INH-PZA-EMB) did not [89].

Two kinds of persisters are known [74]: (i) Class I, rare, generated in a replicating population, formed continuously and in a purely stochastic manner. They are bacilli phenotypically tolerant to different antibiotics by different mechanisms, and it is likely that the overall population can be killed by drug combinations; (ii) Class II, abundant, involving almost all of the cells in a population, e.g., in the stationary phase, hypoxic conditions, nutrient starvation. Growth arrest is associated with resistance to a large number of drugs, and it is likely that new kinds of antibiotics are necessary to overcome these cells [74].

Dormancy is not necessary or sufficient for Mtb persistence, indicating that persistence is a phenomenon more complex than dormancy, and that additional characteristics are needed to define the persister phenotypes, which depends on the NR model used [90]. A poor correlation was found between the transcriptomes of class I persisters enriched by cycloserine [91] and class II persisters obtained under hypoxia, the stationary phase or nutrient starvation [74]. On the other hand, persister diversity is expected also from the different host environments in which these specialized cells live, ranging from the intracellular location in the phagosomes to extracellular life in the caseum. In BDQ-treated guinea pigs, persisting bacilli where located in the acellular rim of necrotic lesions, morphologically similar to human TB lung lesions [92].

The state of non-replication is associated with phenotypic drug-tolerance, but different stresses may induce phenotypically different bacilli. Few compounds were dual active molecules with bactericidal activity against both replicating and NR Mtb. They included RIF, BDQ, PRT and MFX, which target RNA polymerase, ATP synthase, cell wall synthesis/cell respiration, and DNA gyrase, respectively [24,31,74]. In BALB/c mice, persisters were eradicated by regimens containing high-dose RIF and BDQ [93,94]. In BALB/c mice, C3HeB/FeJ caseum-forming mice and athymic nude mice, PRT contributed significantly to the efficacy of BDQ-containing regimens, with either LZN (BPaL regimen) or MFX and PZA (BPaMZ regimen) [95].

Interestingly, RIF-resistant or MXF resistant mutants carrying mutations in *rpoB* or *gyrA* genes emerged at high frequency from the persistent phase of Mtb cells exposed to RIF for prolonged periods. These cells carried elevated levels of the hydroxyl radical, which inflicted genome-wide mutations facilitating resistance to the same, or another, antibiotic [38,39]. In consideration of the long TB therapy, these observations may have clinical significance in the emergence of drug-resistant mutants if local monotherapy occurs in patients who do not correctly take multi-drug TB therapy.

In this view, it was postulated that persisters behave as an evolutionary reservoir from which drug-resistant mutants can emerge [22].

Thus, targeting NR persisters could reduce the duration of antibiotic treatment and rate of post-treatment relapse [74,96]. Researches aimed at better understanding the relationship between persistence and resistance, and at finding novel drug combinations for killing both AR and NR bacilli, will provide new strategies to shorten TB therapy.
