Mab subsp. abscessus CIP 104536^T S and R morphotypes were kindly provided by Dr. Laurent Kremer (CNRS, IRIM, Universite’ de Montpellier, Montpellier, France). Mab subsp. bolletii CIP108541^T and Mab subsp. massiliense CIP108297^T were purchased from the Collection de l’Institut Pasteur (CIP, Paris, France). Clinical isolates were obtained from the Korea Mycobacterium Resource Center (KMRC, Osong, Korea). AMK and CFX resistant strains used in this study were derived from a previous study (Kim et al., 2017). Mab strains were grown at 37°C in a Middlebrook 7H9 culture medium (Difco), supplemented with 10% albumin-dextrose-catalase (ADC, Difco) and 0.05% Tween-80 (Sigma). For the CFU determination, bacteria was plated in a Middlebrook 7H10 solid culture medium containing 0.5% glycerol and 10% OADC (Difco). In order to evaluate the MIC and drug-drug interaction, the bacteria was tested in a cation-adjusted Mueller–Hinton (CAMH) medium (Sigma, St. Louis, MO, United States) supplemented with 20 mg/L calcium chloride (Sigma, St. Louis, MO, United States) and 10 mg/L magnesium chloride (Sigma, St. Louis, MO, United States). To induce the zebrafish infection, a recombinant Mab CIP 104536^T R morphotype that was carrying a pMV262-mWasabi, that was prepared previously, was used (Kim et al., 2019). All cultures were grown at 37°C while shaking at 180 rpm. CLA, RFB, AMK, and CFX were purchased from Sigma-Aldrich (St. Louis, MO, United States). OMD and BDQ were purchased from Adooq Bioscience (Irvine, CA, United States).

Determination of compound interactions using a REMA checkerboard assay and evaluation of compound interactions using CFU determination

For each drug’s MIC determination, a REMA was performed, as described previously (Hanh et al., 2020a). Furthermore, checkerboard assay using resazurin was performed in a similar manner to that described for Mab by Cheng et al.,with minor modifications (Cheng et al., 2019). The checkerboard method was used to evaluate the antibacterial ability of the two antibacterial drugs. 1 µL of the two-fold serial dilutions of each test compound (starting from 8 × the MIC₅₀) was prepared in a well of a 96-well flat, clear bottom, white microplate (98 µL per well) (Corning, Baltimore, MD, United States). Bacterial stocks of Mab subsp. abscessus CIP 104536^T from the exponential-phase cultures were eluted to an optical density measuring 600 nm (OD₆₀₀) of 0.0025 and added to the plates to obtain a total volume of 100 µL. Each plate was then incubated for 5 days at 30°C, before the addition of resazurin [0.025% (wt/vol) to 1/10 of well volume] as described previously (Cheng et al., 2019). After overnight incubation, fluorescence was measured using a spectraMax® M3 Multi-Mode Microplate Reader (Molecular Devices, Sunnyvale, CA, USA) with excitation at 560 nm and emission at 590 nm.

To evaluate compound interactions, fractional inhibitory concentrations (FICs) were calculated using the following formula: FIC (X + Y) = [MIC of compound X in combination with Y]/[MIC of X alone]. The fractional inhibitory index (ΣFIC) is the sum of the FIC of compound X and the FIC of compound Y. Synergy was defined by ΣFIC values of ≤0.5, antagonism by ΣFIC values> 4.0, and values in between correspond to additivity (Odds, 2003). The isobologram curves showing the result of the interaction of the two antibacterial agents from the MICs for the antibacterial agents when used alone, or in combination, were constructed using GraphPad Prism software (version 6.05; San Diego, CA, United States). To detect the bacterial viability, bacteria was first incubated in the presence of combinations of the compounds at their respective MICs before they were then plated on solid Middlebrook 7H10 mediums (Difco). CFU counts were determined after 3 days of incubation at 37°C.

All ZF experiments were approved by the Animal Research Ethics Committee of Gyeongsang National University (Project identification code: GNU-190325-E0014, Approval date: Mar 25, 2019).

Mab CIP 104536^T R morphotype, harboring mWasabi, was selected under the pressure of kanamycin 50 mg/L. The infection stock was prepared as described previously (Hanh et al., 2020b). Infection stock was then diluted with PBST (Phosphate-Buffered Saline with 0.05% Tween 80) and re-suspended in Phenol Red 0.085%. The zebrafish larvae at 30–48 h post-fertilization were dechorionated and anesthetized with 270 mg/L tricaine at room temperature. Around 3 nL of MabR-mWasabi (400 CFU) was injected via the caudal veins using a Tritech Research Digital microINJECTOR (Tritech research, model MINJ-D). The infected larvae were transferred into 96-well plates (2 fish per well containing 200 µL water) and exposed to various drug combinations. CLA (3.1 µM) was combined with other anti-Mab agents (OMD 6.3 µM, BDQ 3.1 µM, AMK 12.5 µM, RFB 6.3 µM, and CFX 6.3 µM). The fish water and compounds were renewed once daily. The ZF larvae treated with DMSO as a vehicle were used as a negative control.

ZF in vivo drug efficacy was assessed as described previously (Hanh et al., 2020a). Briefly, the in vivo anti-Mab effect of each drug combination was determined through GFP dissemination, counts of CFU, and the survival curve. GFP quantification was accessed by capturing the MabR-mWasabi evolution inside the infected larvae at 5 days post-infection using an ImageXpress Pico Automated Cell Imaging System (Molecular Devices, Sunnyvale, CA, United States). For the quantification of bacterial load, a group of 20 infected embryos (5 dpi) was collected and individually homogenized in 2% Triton X-100–PBST using a handheld homogenizer (D1000; Benchmark Scientific, Sayreville, NJ, United States). Serial 10-fold dilutions of the suspension were plated out on Middlebrook 7H10 solid culture mediums containing 50 μg/ml kanamycin and BBL™ MGIT™ Mycobacteria Growth Indicator PANTA (polmyxin B, amphotericin B, nalidixic acid, trimethoprim, and azlocillin; Becton Dickinson, Franklin Lakes, NJ, United States), and then incubated for 3–5 days at 37°C to enumerate the CFU. The number of dead embryos (no heartbeat) was recorded daily, for 13 days, to determine the survival curve. The CFU quantification and survival curve were plotted by Prism using the method from Kaplan and Meier and the log-rank (Mantel–Cox) test, respectively, to compare the difference between untreated control and treated embryos.

To find the best combination with CLA, various drugs were included in this experiment. First, the MIC₅₀ value of each individual compound was determined by Resazurin Microtiter Assay (REMA). MIC₅₀ value was defined as the minimum inhibitory concentration (MIC) required to inhibit 50% growth of the organism. The MIC values of all tested compounds are presented in Table 1. CLA showed favorable activity against Mab (MIC₅₀ = 5 µM). Second, drug-drug interactions were evaluated to find the best combination with CLA, using mid-log phase cells of Mab utilizing a checkerboard assay. CLA concentrations ranging from 0 to 39.6 µM (8 points) were prepared in 96 well plates through 2-fold serial dilution, and the MIC₅₀ values of CLA were placed at the middle of the concentration range. This gradient CLA concentration was used to test interactions with five different anti-Mab drugs such as OMD, AMK, rifabutin (RFB), BDQ, and CFX in various drug concentrations, based on 2-fold serial dilution. The interactions were interpreted using a fractional inhibitory concentration (FIC) index for each combination (Table 1). The experiment was repeated three times and the combination effect was consistent across the replicated experiments. As shown in Figure 1, the concentrations of CLA and OMD required for this synergistic effect were much lower than their MIC alone. For example, one-half the MIC₅₀ of CLA (pink) added to one-half the MIC₅₀ of OMD (pink) did not result in a resazurin color change in the dye from blue-purple to pink, which indicates the inhibition of Mab growth (Figures 1A,B). Furthermore, one-quarter the MIC₅₀ of CLA added to one-quarter the MIC₅₀ of OMD also prevented resazurin turnover. Synergy has traditionally been defined with a FIC index of 0.5 or less (Braga et al., 2005). Thus, the CLA plus OMD combination effect is synergistic against Mab. Interestingly, the CLA-BDQ combination also showed a synergistic effect (FIC = 0.5) against Mab, although the FIC index was higher than the CLA-OMD combination (Table 1; Supplementary Figures S1A, S1B). The CLA with AMK, RFB, and CFX combinations showed no synergistic antimicrobial effects, with a FIC value of over 0.5. CLA showed an additive effect with AMK (Figures 1C,D), RFB (Figures 1E,F), and CFX (Supplementary Figures 1C, 1D) against Mab, with FIC index values of 0.7–1.4. This means each compound did not interact in a direct way without affecting the other (Table 1). No antagonistic interactions were found between CLA and the compounds tested.

To confirm the synergistic effect against Mab, a traditional CFU (colony forming unit) determination assay was conducted. As shown in Figure 2A, the obtained results were consistent with the results from the checkerboard method. The CFU determination assay confirmed that combinations of CLA with OMD showed a clear synergistic effect leading to a significant reduction in bacterial numbers on the agar plates. The results show that the combination of 2.47 μM of CLA (one-half the MIC) and 0.85 µM OMD (one-half the MIC) had clear growth inhibitory activity (6.4 log₁₀ cfu/mL reduction), compared to the activity of the untreated DMSO control on day 7. In addition, this combination also showed at least a 3 log₁₀ cfu/mL reduction compared to the single CLA and OMD samples respectively. Based on the definition, bactericidal activity was defined as a reduction of at least ≥3 log₁₀ of the total count of CFU/mL in the original inoculum. Therefore, the CLA plus OMD combination is shown to be bactericidal against Mab (Kragh et al., 2021). Again, the CLA-BDQ combination also showed significant bacterial reduction, as similar with the checkerboard assay. However, the CLA-BDQ combination showed less than a 3 log₁₀ cfu/mL reduction compared to the single BDQ. Thus, this combination was considered to be bacteriostatic against Mab (Supplementary Figure S2A). Conversely, CLA in combination with AMK, RFB, and CFX acted additively, with the combinations giving similar inhibition of bacterial viability to the single agents (Figure 2A; Supplementary Figures S2B, S2C). Furthermore, we tested CLA-OMD effectiveness against Mab subspecies and clinical isolates that have different morphotypes, including AMK and CFX laboratory induced resistant strains that were generated in a previous study (Kim et al., 2017). Table 2 shows the MIC of CLA and OMD alone, and the FIC values in combination, against 3 Mab subspecies, 7 clinical isolates (6 Mab subsp. abscessus and 1 Mab subsp. massiliense), and AMK and CFX resistant strains. Synergism was found in 100% of the strains tested. Three different Mab R morphotypes (Mab subsp. abscessus CIP104536, KMRC 00136-61040, and KMRC 00200-61202) also showed a synergistic effect (FIC index less than 0.5).

Furthermore, we verified the CLA-OMD combination effectiveness in Mab-infected ZF. To do this, the dilution of each drug evaluated was set as the concentration that resulted in a reduction of 1 log₁₀ CFU in infected ZF survival, compared to the untreated control, by serial drug dilution at 5 days post-infection (dpi) (data not shown). From this drug serial dilution, CLA (3.1 µM), BDQ (3.1 µM), AMK (12.5 µM), OMD (6.3 µM), RFB (6.3 µM), and CFX (6.3 µM) were determined as the drug concentrations that yielded approximately a 1 log₁₀ CFU reduction on an agar plate, when compared to the untreated control (Figure 3A). For the next step, an in vivo combination drug efficacy test was performed with these selected concentrations of each drug. CLA was used as the anchor drug and was separately paired with BDQ, AMK, OMD, RFB, and CFX. As shown in the survival curve (Figure 3B), double therapy with CLA (3.1 µM) and OMD (6.3 µM) yielded a significantly lower mortality rate than the other combinations. The CLA plus OMD combination led to a 30.5% mortality rate at 13 days after treatment. In contrast, the conventional pairing of CLA (3.1 µM) plus AMK (12.5 µM), or CFX (6.3 µM), showed a higher mortality rate (60.5 and 90% of Mab-infected ZF at 13 dpi, respectively). The combination of CLA (3.1 µM) plus BDQ (3.1 µM) was also not effective. It showed a 10% survival rate for infected ZF at 13 dpi. Furthermore, the combination between CLA and RFB also showed almost no synergistic effect. The Mab infected and untreated ZF group 100% died and the non-infected group 100% survived after 13 days.

The bacterial burden in the ZF was measured by conventional CFU counts and fluorescence microscopy after different combinations of treatments. To determine whether each combination effectively reduced the bacterial burden in the ZF, bacterial survival was compared with the CLA (3.1 µM) and OMD (6.3 µM) treatments and the other CLA combinations (BDQ, AMK, RFB, and CFX), including the non-treated DMSO control and non-infected ZF. To do this, each infected and treated ZF was crushed and sampled, and the number of bacteria was enumerated on a 7H10 agar plate. Figure 3C shows that Mab replicated inside the hosts, and the CFU showed significant differences between the groups. The lowest bacterial CFU per ZF was observed in the presence of CLA (3.1 µM) and OMD (6.3 µM) as expected (Figure 3C). This combination showed around a 2.8 log₁₀ reduction compared with the non-treated DMSO control group 5 days after injection. This result was consistent with the observed survival rate in infected ZF (Figure 3B). To confirm the colonization of Mab inside the ZF bodies under treatment with different combinations, we also used a mWasabi green fluorescent protein (GFP) labeled Mab strain. The GFP levels were observed using an ImageXpress® Pico Automated Cell Imaging System on anaesthetized ZF. It allowed us to measure the progression of GFP labelled bacterial colonization following combination treatment. ZF treated with CLA (3.1 µM) plus OMD (6.3 µM) were compared with those treated with other combinations and control ZFs. GFP labelled Mab dissemination was observed in the brain and yolk in the non-treated DMSO control (Figure 3D). However, ZFs treated with CLA (3.1 µM) and OMD (6.3 µM) showed almost no GFP fluorescence (Figure 3D). GFP signals in the brain area were still observed in other CLA combinations, although the GFP signal in the ZF yolks disappeared. These results were consistent with the survival curves and the CFU determination (Figures 3B,C). Therefore, these results indicate that the combination of CLA and OMD significantly inhibited Mab growth in the ZF bodies and, consequently, extended the lifespan of the infected ZF.