low pH, M. tuberculosis slows or halts replication and becomes
sensitive to pyrazinamide.
Pyrazinamide is effective in treating TB. The simplest nonreplicating model, “Loebel,” starves M. tuberculosis by prolonged
incubation in a nutrient-free buffer such as PBS [13–15]. Starved
M. tuberculosis are resistant to most known drugs used to treat
M. tuberculosis, although they can still be killed by rifampicin, the
phenothiazines, trifluoperazine, and chlorpromazine [16–18]. The
Wayne model deprives an initially replicating M. tuberculosis culture
of O 2 in a gradual manner that allows the cells to survive severe
hypoxia [19–21]. M. tuberculosis rendered severely hypoxic in the
Wayne model were resistant to isoniazid, rifampicin, and ciprofloxacin and sensitive to metronidazole and the combination of meropenem and clavulanate [20, 22]. Given metronidazole’s activity
on hypoxic M. tuberculosis, it was predicted that metronidazole
would be effective against M. tuberculosis in guinea pigs, rabbits,
and nonhuman primates, because immunostaining for pimonidazole adducts (a reaction dependent on the compound’s reduction
by an O 2 -inhibited enzyme) indicated that their necrotic granulomas were hypoxic [2]. In contrast, no activity was anticipated in
mice, as their nonnecrotizing lesions were not hypoxic [23]. As
predicted, metronidazole lacked activity against M. tuberculosis in
mice [24]. In rabbits, metronidazole killed Mycobacterium bovis
BCG, [23] and in cynomolgus macaques, it prevented antiTNFdependent relapse of latent TB [25]. However, metronidazole
failed to kill M. tuberculosis in guinea pigs [26] and was ineffective
in humans [27].
The case for conducting HTS against nonreplicating
M. tuberculosis was made over a decade ago, [28] but development
of an assay was problematic. The low levels of ATP in nonreplicating M. tuberculosis [29] make it difficult to use ATP-based readouts
as a measure of viability [12, 30]. An early innovation for a nonreplicating HTS was the addition of a short outgrowth phase (28 h)
after the nonreplicating phase to improve the luciferase signal
[30]. Building on this concept, we developed a multistress nonreplicating assay for HTS by splitting the assay into two phases—a
nonreplicating phase and an outgrowth phase—and extending the
duration of the latter (Fig. 1). In the first phase, a logarithmically
replicating culture of M. tuberculosis is washed with PBS, dispensed
into the nonreplicating conditions described in Table 1, which are
titrated to maintain colony-forming units (CFU) at a near-constant
level, and incubated in the presence of DMSO as a negative control,
rifampin or oxyphenbutazone as positive controls, or test compounds. After 3 days, an aliquot of the contents of each well is
diluted into a new microplate that is cultured in replicationsupporting conditions. M. tuberculosis surviving the drug treatment
grow during the outgrowth phase, while dead cells fail to increase
their biomass. Alternatively, a higher-throughput, less expensive
612
Ben Gold et al.
sensitive to pyrazinamide.
Pyrazinamide is effective in treating TB. The simplest nonreplicating model, “Loebel,” starves M. tuberculosis by prolonged
incubation in a nutrient-free buffer such as PBS [13–15]. Starved
M. tuberculosis are resistant to most known drugs used to treat
M. tuberculosis, although they can still be killed by rifampicin, the
phenothiazines, trifluoperazine, and chlorpromazine [16–18]. The
Wayne model deprives an initially replicating M. tuberculosis culture
of O 2 in a gradual manner that allows the cells to survive severe
hypoxia [19–21]. M. tuberculosis rendered severely hypoxic in the
Wayne model were resistant to isoniazid, rifampicin, and ciprofloxacin and sensitive to metronidazole and the combination of meropenem and clavulanate [20, 22]. Given metronidazole’s activity
on hypoxic M. tuberculosis, it was predicted that metronidazole
would be effective against M. tuberculosis in guinea pigs, rabbits,
and nonhuman primates, because immunostaining for pimonidazole adducts (a reaction dependent on the compound’s reduction
by an O 2 -inhibited enzyme) indicated that their necrotic granulomas were hypoxic [2]. In contrast, no activity was anticipated in
mice, as their nonnecrotizing lesions were not hypoxic [23]. As
predicted, metronidazole lacked activity against M. tuberculosis in
mice [24]. In rabbits, metronidazole killed Mycobacterium bovis
BCG, [23] and in cynomolgus macaques, it prevented antiTNFdependent relapse of latent TB [25]. However, metronidazole
failed to kill M. tuberculosis in guinea pigs [26] and was ineffective
in humans [27].
The case for conducting HTS against nonreplicating
M. tuberculosis was made over a decade ago, [28] but development
of an assay was problematic. The low levels of ATP in nonreplicating M. tuberculosis [29] make it difficult to use ATP-based readouts
as a measure of viability [12, 30]. An early innovation for a nonreplicating HTS was the addition of a short outgrowth phase (28 h)
after the nonreplicating phase to improve the luciferase signal
[30]. Building on this concept, we developed a multistress nonreplicating assay for HTS by splitting the assay into two phases—a
nonreplicating phase and an outgrowth phase—and extending the
duration of the latter (Fig. 1). In the first phase, a logarithmically
replicating culture of M. tuberculosis is washed with PBS, dispensed
into the nonreplicating conditions described in Table 1, which are
titrated to maintain colony-forming units (CFU) at a near-constant
level, and incubated in the presence of DMSO as a negative control,
rifampin or oxyphenbutazone as positive controls, or test compounds. After 3 days, an aliquot of the contents of each well is
diluted into a new microplate that is cultured in replicationsupporting conditions. M. tuberculosis surviving the drug treatment
grow during the outgrowth phase, while dead cells fail to increase
their biomass. Alternatively, a higher-throughput, less expensive
612
Ben Gold et al.