Isoniazid and rifampicin were purchased from Sigma-Aldrich. Dexamethasone, CLR, BDQ, CFZ, MOFX, ofloxacin, RFB, PBTZ169 and pretomanid were purchased from Biochempartner Co., Ltd. LZD was purchased from Ark Pharm, Inc.

M. tuberculosis (H37Rv), M. kansasii (ATCC 12478), M. marinum (ATCC 927), M. gordonae (ATCC 14470), M. scrofulaceum (ATCC 19981), M. avium (ATCC 25291), M. intracellulare (ATCC 13950), M. xenopi (ATCC 19250), M. abscessus (ATCC 19977), M. chelonae (ATCC 14472), M. fortuitum (ATCC 6841), and M. gilvum (ATCC 43909) were obtained from the National Clinical Laboratory on Tuberculosis, Beijing Chest Hospital. The M. tuberculosis and non-tuberculous mycobacterial strains were grown in Middlebrook 7H9 broth (Difco) supplemented with 0.2% (vol/vol) glycerol, 0.05% Tween 80, and 10% (vol/vol) oleic acid-albumin-dextrose-catalase (Becton-Dickinson).

MICs were determined by a microplate-based alamarBlue assay (Xu et al., 2019). Briefly, bacteria (2 × 10⁵ CFU/mL) were added to wells, yielding a final testing volume of 200 μL. The plates were incubated at 37°C. On day 3 (for RGMs) or 7 (for SGMs) of incubation, 20% Tween 80 (12.5 μL) and alamarBlue (20 μL) were added to all wells. After incubation at 37°C for another 24 h, the fluorescence was measured at an excitation wavelength of 530 nm and an emission wavelength of 590 nm. The MIC was defined as the lowest concentration eliciting a reduction in fluorescence of ≥90% relative to the mean fluorescence of replicate drug-free controls. M. tuberculosis H37Rv was used as a drug-susceptible control.

Seventy female BALB/c mice (18 to 20 g) were inoculated intravenously in the tail vein. Dexamethasone (5 mg/kg) was freshly prepared in carboxymethyl cellulose (CMC), and the treatment began 1 week prior to infection, was stopped 6 days after infection, administered daily, and was not administered on the day of infection. To achieve an implantation of 4.0–5.0 log₁₀ CFU in the lungs of a mouse, a primary M. abscessus culture in the exponential phase, with A_600nm of 1.00–1.20, was used to prepare a suspension by diluting to a calculated A_600nm of 0.1 in normal saline. CLR was administered 5 days a week for 3 weeks after stopping dexamethasone. Sacrifices were performed 1 day post infection (D-6), on the day of CLR administration (D0), and then at D7, D14, D21, and D28 post infection.

M. abscessus, M. chelonae, and M. fortuitum mouse models were established as described above. One hundred and eighty mice were randomly divided into four groups and infected with four types of mycobacteria. Five mice infected with different bacteria were humanely killed 1 day after infection and on the day of treatment initiation. The number of bacteria implanted in the lungs and spleen (or kidneys) were determined by plating serial dilutions of tissue homogenates on nutrient 7H10 agar. The remaining 35 mice in each group were randomly allocated among seven groups, including one untreated control group, and six groups were treated with BDQ, CFZ, RFB, MOFX, pretomanid, or PBTZ169 monotherapy.

Treatment was initiated 7 days after infection. For treating the mice, BDQ, CFZ, RFB, MOFX, PBTZ169 and pretomanid were suspended in 0.5% (wt/vol) CMC, whereas MOFX was diluted with normal saline. All antimicrobial agents were given for 28 days (7 times weekly) by gavage, and mice in the control group were administered CMC. The dosages for each treatment, per kilogram of body weight, were 20 mg/kg RFB (Singh et al., 2021), 25 mg/kg BDQ (Zhang et al., 2019), 25 mg/kg CFZ (Baijnath et al., 2015), 25 mg/kg PBTZ169 (Makarov et al., 2014), 100 mg/kg MOFX (Andrejak et al., 2015) and 200 mg/kg pretomanid, identical to the effective dosages of these antimicrobial agents against M. tuberculosis infection of mice. The dose selection of pretomanid was based on our previous experiments (data not shown), in which 100 mg/kg pretomanid (Xu et al., 2019) was ineffective against NTM in mice, so 200 mg/kg was used in this experiment.

Efficacy was assessed based on the lung, spleen, or kidney (mice infected with M. fortuitum) CFU counts. Comparisons to determine significant effects of antibiotic compounds are in reference to the D28 untreated control. Mice were sacrificed 3 days after the last day of treatment to reduce carryover effects. The organs were aseptically removed and homogenized. Suspensions were made up to 1 mL for each organ. At least four serial 10-fold dilutions of the suspension were performed on 0.5% charcoal-containing selective 7H10 plates. The plates were incubated for up to 1 (for RGMs) or 4 (for SGMs) weeks at 37°C before the final CFU counts were determined.

Organ CFU counts were log-transformed before analysis, and mean CFU counts were compared by one-way analysis of variance with Dunnett’ s post hoc test to control for multiple comparisons. The Mann-Whitney test was used to test for significance on nonnormally distributed CFU data. All analyses were performed with GraphPad Prism version 5 (GraphPad, San Diego, CA). A P value of 0.05 was considered significant.

As shown in Table 1 , first-line antitubercular drugs, isoniazid and rifampicin, had limited activities against NTM in vitro, especially against M. avium, M. abscessus, M. chelonae, and M. fortuitum, and they were partly used as controls. Clarithromycin (CLR) is a baseline drug for the treatment of NTM lung disease, and our study showed that CLR had good activities against all NTMs except M. chelonae (24.922 μg/mL).

Of the remaining nine drugs (or compound), PBTZ169 showed good activities against M. kansasii, M. marinum, M. gordonae, M. xenopi, M. fortuitum and M. gilvum (MICs of <0.016 μg/mL), but had no activity against other 5 NTMs (MICs of >32 μg/mL) against the 11 NTMs. Pretomanid exhibited poor in vitro activities against most NTMs we tested, only M. kansasii and M. xenopi were sensitive to pretomanid. Bedaquiline (BDQ) and rifabutin (RFB) had the lowest range of MIC values for the antibiotics we tested. The MICs for BDQ against 11 reference strains ranged from 0.01 to 1.96 μg/mL, which were higher than those for RFB (MICs of <0.001 to 5.86 μg/mL). For the fluoroquinolones, the activity of moxifloxacin (MOFX) against NTM was superior to that of ofloxacin. Furthermore, clofazimine (CFZ) and linezolid (LZD) had different activities against NTM in RGMs and SGMs, with MICs of 0.12 to 4.51 μg/mL against the RGMs and 0.39 to 5.81 μg/mL against the SGMs, respectively. CFZ was moderately active against all four RGMs (MICs of 1.92to 12.64μg/mL), whereas LZD was ineffective against all three RGMs, except M. gilvum (MIC of 0.52 μg/mL).

BDQ, CFZ, RFB, and MOFX were the four most active of the 11 antibiotics, so their in vitro activities against three common NTM clinical strains, M. avium, M. abscessus, and M. fortuitum, were tested. Because we did not collect clinical strains of M. chelonae, no evaluation was performed. Although the NTMs were insensitive to PBTZ169 and pretomanid in vitro, these compounds were also included in subsequent studies. The results suggested that BDQ, CFZ, RFB, and MOFX also had good in vitro activities against M. avium, M. abscessus, and M. fortuitum clinical strains ( Table 2 and Supplementary Table 1).

To assess the efficacy of antibiotics, we established murine models of M. avium, M. abscessus, M. chelonae, and M. fortuitum disease. Although the anti-NTM activities of PBTZ169 and pretomanid were poor in vitro, they significantly reduced the M. marinum load in zebrafish (Makarov et al., 2014; Dalton et al., 2017); thus, we evaluated the anti-NTM activities of BDQ, CFZ, RFB, MOFX, PBTZ169, and pretomanid in mice.

The bacterial load of M. avium in mouse organs continued to rise throughout the experimental period ( Figure 2 , Table 3 ). At 1 day post infection, all mice were implanted with 5–6 log₁₀ CFU of M. avium in the lungs and spleen. At the end of treatment, the intrapulmonary bacterial load was ~7 log₁₀ CFU and the bacterial load in the spleen was ~8 log₁₀ CFU. PBTZ169 and MOFX showed weak bacteriostatic activities against M. avium, reducing the bacterial load by <1 log₁₀ CFU. In addition, pretomanid and RFB did not have an inhibitory effect on M. avium. BDQ and CFZ showed excellent antimicrobial activities against M. avium in the lungs and spleen, and they reduced the bacterial load in the lungs by 4.76 and 3.53 log₁₀ CFU and in the spleen by 3.40 and 4.42 log₁₀ CFU, respectively.

In the mice infected with M. abscessus, as with previous results, a high bacterial load was maintained in the lungs throughout the infection and treatment periods. However, the bacterial load in the spleen showed a continuous decrease; thus, the antibacterial effect of the drug present in the spleen was partly attributed the clearance of mycobacteria by the mice’s immune system ( Figure 3 ; Table 4 ). Throughout the experiment, three mice (5.5%) developed characteristic spinning disease, in which the mice held their heads to one side and often shook and twitched their heads, and when suspended by their tails, they rotated vigorously. In contrast to the in vitro sensitization results, both PBTZ169 and pretomanid had strong antimicrobial activities in the lungs (3.33 and 3.12 log₁₀ CFU reduction, respectively), and pretomanid had strong antimicrobial activity (2.30 log₁₀ CFU reduction) in the spleen too, whereas PBTZ169 reduced the bacterial load by only 1.49 log₁₀ CFU. BDQ showed good activities against M. abscessus in vitro and in vivo in the lungs (3.09 log₁₀ CFU reduction) and in the spleen (2.56 log₁₀ CFU reduction). The antibacterial activity of CFZ and MOFX in the lungs (2.30 and 2.25 log₁₀ CFU reduction) was better than that in the spleen (1.37 and 1.20 log₁₀ CFU reduction). Although M. abscessus was sensitive to the RFB in vitro, RFB had no antimicrobial activity in mouse lungs or spleen.

Unlike M. abscessus, M. chelonae remained at high levels in the lungs and spleen throughout the infection and treatment periods ( Figure 4 ; Table 5 ). Throughout the experiment, PBTZ169 and pretomanid were equally ineffective against M. chelonae in vitro, but unlike the M. abscessus infection group, pretomanid had little in vivo antimicrobial activity, whereas PBTZ169 showed good antimicrobial activities in both lungs and spleen (2.29 and 2.24 log₁₀ CFU reduction). The activities of BDQ and CFZ against M. chelonae in mice in the lungs (3.40 and 3.68 log₁₀ CFU reduction) and spleen (2.07 and 2.56 log₁₀ CFU reduction) were similar to those for M. abscessus. However, unlike the M. abscessus group, RFB had better activities against M. chelonae (3.43 log₁₀ CFU reduction in the lungs and 1.25 log₁₀ CFU reduction in the spleen). MOFX was active only in the lungs (1.41 log₁₀ CFU reduction), and had weak activity in the spleen.

Almost half of the mice infected with M. fortuitum developed rotational disease (27/55, 49.1%), which was the highest of the four animal models we established. The bacterial load in the spleen was 4.49 ± 0.19 log₁₀ CFU/spleen 1 day after M. fortuitum infection, whereas it decreased to 3.44 ± 0.13 log₁₀ CFU/spleen on the day of treatment (data not shown). This was consistent with previous reports that M. fortuitum in the spleen in mice decreases over time until complete clearance (Parti et al., 2005). The bacterial load in the lungs was 4.56 ± 0.06 log₁₀ CFU/lung on day 1 post infection, remained stable at 4.51 ± 0.19 log₁₀ CFU/lung on the day of treatment, and decreased to 1.97 ± 0.20 log₁₀ CFU/lung after 28 days of treatment, also consistent with the trend reported in the literature, but with a greater decrease in CFU. Although there was a significant decrease in intrapulmonary bacterial load in mice with obvious immune clearance involved, BDQ and PBTZ169 had moderate activities against M. fortuitum in mouse lungs, whereas RFB, CFZ, and MOFX were more active ( Figure 5A ; Table 6 ). Next, we analyzed the antibacterial activities of the drugs in mouse kidneys ( Figure 5B ; Table 6 ). Pretomanid showed only weak activity (0.88 log₁₀ CFU reduction), whereas PBTZ169 exhibited moderately strong activity (1.62 log₁₀ CFU). MOFX was the most active against M. fortuitum, whitch completely cleared the bacterial load from the kidneys. BDQ and RFB showed consistent activities in vitro and in vivo, and they exhibited bactericidal activities in the kidneys (2.41 and 3.15 log₁₀ CFU reduction). CFZ cleared M. fortuitum from the lungs, but only inhibited the bacteria in the kidneys.