Pathogen Box screening for hit identification against Mycobacterium abscessus

Mycobacterium abscessus is a rapidly growing life-threatening mycobacterium with multiple drug-resistance mechanisms. However, there is no official regimen for M. abscessus therapy. In this study, we screened the Pathogen Box, which contains 400 drug-like molecules active against neglected diseases, to identify active molecules targeting Mycobacterium abscessus using resazurin live/dead assays. In this screening assay, the Z-factor was 0.7, as an indicator of the statistical confidence of the assay. A cut-off of 80% growth inhibition in the screening resulted in the identification of four different compounds at a single concentration (20 μM). Dose-response curves identified three different hit candidates, i.e., MMV688508, MMV688844, and MMV688845, which generated good inhibitory curves. All hit candidates were expected to have different molecular targets. Among them, MMV688844 showed the best minimum inhibitory concentration value for not only wild-type M. abscessus but also for nine different R and S morphotype clinical isolates. Thus, we found that MMV688844, identified from the Pathogen Box screen, may be a promising candidate in the M. abscessus drug discovery pipeline.


Introduction
Mycobacterium abscessus is a potentially life-threatening, pathogenic, rapidly growing mycobacterium (RGM) with multiple intrinsic and extrinsic drug-resistance mechanisms [1]. For example, M. abscessus contains a thick and waxy mycobacterial cell envelope, similar to other mycobacteria, preventing the penetration of toxic drugs and consequently resulting in poor drug effectiveness. In addition, the M. abscessus genome contains many genes involved in drug-efflux systems, such as members of the major facilitator family, ABC transporters, and MmpL proteins [2]. M. abscessus also possesses enzymes possibly involved in natural resistance via the modification and inactivation of antibiotics, such as rifampicin ADP-ribosyltransferase and a mono-oxygenase that may be involved in resistance to rifampicin [2]. An inducible erythromycin ribosome methyltransferase erm (41) (MAB_2997), which attenuates clarithromycin and erythromycin activity, was also recently discovered [3]. Extrinsically, M. abscessus is equipped with acquired antibiotic resistance through spontaneous mutations of molecular

Assay validation and hit screen
The assay was validated using the Z-factor. For this, a series of negative and positive controls were measured in 7H9 medium supplemented with 10% (v/v) albumin-dextrose-saline (ADS) enrichment, 0.2% (v/v) glycerol, and 0.05% (v/v) Tween 80, which was inoculated with M. abscessus (5 × 10 5 bacteria per well). A total of 1 μL of 10 μM clarithromycin in DMSO solution was used as a positive control, and 1% of the DMSO solution was used as a negative control. The plate was incubated for 40 h at 37˚C. After adding resazurin at 0.02% (wt/vol) in 40 μL per well and incubation for 12 h at 37˚C, the plates were analyzed using a SpectraMax1 M3 Multi-Mode Microplate Reader (Molecular Devices, CA) at an excitation wavelength of 560 nm and emission wavelength of 590 nm, at which the color of the dye changed from blue to pink and the fluorescence in negative control cultures reached 800 relative fluorescence units (RFU). Resazurin is a common marker used to distinguish between live and dead Mycobacterium sp. based on the degree of metabolism. Metabolically active bacteria can reduce blue non-fluorescent oxidized resazurin within an environment of viable cells to resorufin, which is pink and fluorescent [15,16]. The assay was performed as described above in two separate experiments. To validate the degree of separation, the Z-factor and the percent inhibition of the positive and negative controls were determined using the formula: where σ p and σ n are the standard deviations of the positive and negative controls, respectively, and μ p and μ n are the corresponding mean values. A Z-factor between 0.5 and 1.0 indicates an excellent assay and statistically reliable separation between the positive and negative controls. The Pathogen Box was screened against M. abscessus at 20 μM. M. abscessus was seeded in 96-well plates (5 × 10 5 bacteria per well) and was exposed to each compound for 40 h at 37˚C. Medium with bacteria alone was used in the negative control wells. Resazurin was added to each well (0.02%, wt/vol), and the plates were analyzed at an excitation of 560 nm and emission of 590 nm. The percentage of growth inhibition was calculated using the RFU value, as described previously [17].

Drug susceptibility testing (DST)
The MICs of the hit compounds were determined using twofold serial dilution assays. For MICs against M. abscessus clinical isolates, nine clinical strains were provided from the Korea Mycobacterium Resource Center (KMRC). Briefly, serial twofold dilutions of the hits were prepared in 7H9 supplemented with ADC, glycerol (0.2%, vol/vol), and Tween 80 (0.05%, vol/vol) in 96-well clear microplates (SPL Inc.) to obtain concentration ranges from 200 to 0.39 μM. M. abscessus WT and clinical isolates were then added to each well at a final concentration of 5 × 10 5 bacteria per well. Microplates were incubated at 37˚C for 40 h. A change in color from blue to pink was used as an indicator of bacterial growth. Concentrations required to inhibit bacterial growth by 50% (IC50s) were determined by fitting the curves with a sigmoidal dose-response using Graph-Pad Prism software (version 6.05). Amikacin was used as a reference compound.

Time-kill assays
An early exponential phase mycobacterial culture (10 7 cfu/mL) was prepared in 30 mL of 7H9 supplemented with ADC, glycerol (0.2%, vol/vol), and Tween 80 (0.05%, vol/vol). For time-kill kinetics assay, cefoxitin and MMV688844 were added at final concentrations of 2.0 and 0.2 mg/mL, respectively, and then serial twofold dilutions of each sample were prepared using the mycobacterial culture. All cultures were grown in Corning1 125-mL Polycarbonate Erlenmeyer Flasks with Vent Caps (Product #431143) at 30˚C under shaking conditions (100 rpm) as described previously [18].

Results and discussion
In our study, we identified effective compounds using a screening assay with M. abscessus grown in the mid-log phase. As shown in S1 Fig., we investigated the growth curve of M. abscessus to determine the optimal time point for screening. After 28 h of culture, the mid-log phase was observed; this time point was then used for all experiments. For screening, we used resazurin reduction assays. In a previous report, resazurin reduction demonstrated complete correlation with the MICs obtained by cfu assay, especially for M. tuberculosis, M. bovis BCG, and M. smegmatis in a safe, reliable, easy and cost-effective manner [16]. Therefore, we employed the resazurin reduction assay to assess the cell viability of the rapidly growing mycobacterium, M. abscessus.
The quality of the screen was evaluated using the Z-factor based on the percent inhibition against M. abscessus between the 1.0% DMSO and 10 μM clarithromycin-treated groups as the negative and positive controls, respectively. Fig 1 shows the scatter-plot distribution of the percent inhibition for 1.0% DMSO and 10 μM clarithromycin. The average Z-factor between the 1.0% DMSO negative control and 10 μM clarithromycin-treated positive control in the 96-well test plates was 0.7, indicating that the assay could reliably separate positive and negative controls. These findings supported the feasibility of our drug screening assay for use in M. abscessus screening. The screen was conducted using a one-point concentration (20 μM) of the test compounds. Growth inhibition of at least 80% was defined as the cut-off, yielding four hits (1.0% hit rate; MMV687146, MMV688508, MMV688844, and MMV688845) from the 400 compounds in the Pathogen Box library (Fig 2). As expected, reference compounds that were already included in the Pathogen Box, such as bedaquiline, clofazimine, linezolid, levofloxacin, ofloxacin, rifampicin, radezolid, sutezolid, and auranofin, showed strong inhibition at 20 μM. Other reference compounds that were effective against neglected diseases, such as pentamidine, mefloquine, benznidazole, and miltefosine, showed no activity against M. abscessus in this screen. Therefore, we concluded that our screen exhibited confidence for the identification of hits from the library.
The ability of the hits from the screen to inhibit the growth of M. abscessus was confirmed using a dose-response experiment with resupplied compounds provided by MMV. The MIC 50 values were 1.4, 15, 2.6, and 9.3 μM for MMV687146, MMV688508, MMV688844, and MMV-688845, respectively. Fig 3 shows the dose-dependent inhibitory effects of hit compounds with regard to their activities against M. abscessus. Of the four compounds with confirmed hits, only three, i.e., MMV688508, MMV688844, and MMV688845 (see chemical structures in Fig 3), displayed dose-dependent activity against M. abscessus. Although MMV687146 showed the most potent MIC 50 , this hit failed to reduce the RFU value at a high concentration, in contrast to the other hits. MMV687146 exhibited an RFU of more than 400 at the highest concentration in triplicate experiments. However, active compounds, such as MMV688508, MMV688844, and MMV688845, showed complete bacterial growth inhibition at the highest concentration. We assumed that MMV687146 failed to kill or inhibit bacterial survival completely at a high concentration, although this compound showed the best MIC 50 value. Thus, we decided to omit MMV687146 from our hit list. Amikacin was used as a positive control (MIC 50 = 9.8 μM).
Of our selected hits, MMV688508 is an intermediate in the synthesis of radezolid, a novel oxazolidinone antibiotic used for the treatment of multidrug-resistant (MDR) infections; this compound has been assessed in two phase II clinical trials [19]. Additionally, linezolid, the first oxazolidinone antibacterial agent, has been applied to the treatment of NTMs, including M. abscessus [20,21], and has shown effectiveness in the treatment of chronic and extensively drugresistant tuberculosis [22]. However, the efficacy of linezolid against NTM still varies among different derivatives, and the clinical use of linezolid in patients with NTM can result in adverse events, such as myelosuppression, in the patient [23]. Thus, MMV688508 may have applications in the treatment of antibiotic-resistant M. abscessus without side effects, meeting the currently unmet need for safer oxazolidinone agents for the treatment of M. abscessus. Radezolid was also effective against M. abscessus in our screen as a reference compound (MMV688327). With regard to molecular targeting, known target genes rrl (encoding 23S rRNA) and rplC (encoding 50S ribosomal protein L3) have been reported to have acquired resistance to oxazolidinones [24]. MMV688508 showed an MIC of 15 μM in our dose-response experiment.
Second, MMV688844 (TCMDC-143649) has been previously released from GSK's Tres Cantos Antimycobacterial Set (TCAMS-TB) and showed activity against M. bovis BCG in replicating and non-replicating M. tuberculosis. To evaluate the possible mechanisms of action and molecular targets of MMV688844, predictive computational biology algorithms with structural similarity and GSK historical biological assay data were used, and the target was predicted to be an ABC transporter (Rv0194) [25]. In our study, MMV688844 showed the most potent inhibitory activity, with an MIC 50 as low as 2.6 μM.
Lastly, MMV688845 was found to have efficacy against tuberculosis, as numbered in the fueling open-source library (GSK1729177A). These compounds are potent non-cytotoxic H37Rv hits [26]. In addition, GSK1729177A was previously shown to have MIC 90 values of 50 μM in THP-1 cells and 6.6 μM in H37Rv [27]. In our study, the MIC 50 of MMV688845 was 9.3 μM against M. abscessus.
M. abscessus isolates were collected from the Korea Mycobacterium Resource Center (KMRC). We further evaluated the potency of MMV688844 against nine clinical isolates of M. abscessus, including R and S colony morphotypes. As shown in Fig 4, MMV688844 was equally effective against eight M. abscessus clinical isolates. MMV688844 even strongly inhibited R-type M. abscessus, which tends to be much more virulent and is involved in the chronic colonization of CF airways. Thus, optimization of MMV688844 may have the potential to deliver a suitable candidate for preclinical trials.
In order to determine whether MMV688844 is bactericidal or bacteriostatic, M. abscessus was treated with amikacin and MMV688844 using time-kill kinetics. Fig 5 shows the mycobacterial inhibitory pattern of M. abscessus at different concentrations of the tested antibiotics on different days. In the experiments with cefoxitin, we observed that less than 0.25× MIC did not exhibit an effective killing effect at up to 96 h (Fig 5A) However, cefoxitin killing was observed at  concentrations above 1× MIC. Representatively, 2× MIC showed a 2.5 log reduction in comparison with the growth control that was not treated with cefoxitin at 96 h. In addition, no bacterial regrowth was observed even at 1× MIC. In the case of MMV688844, no significant inhibitory effect was observed below 4× MIC (Fig 5B). However, there was a smaller decline in bacterial cfu/ mL above 8× MIC in comparison with cefoxitin. For example, MMV688844 showed a 1.8 log reduction compared to the growth control at 16× MIC. Inhibition appeared to be maximized at 16× MIC, and regrowth was observed at 96 h for 8× MIC. However, bacterial regrowth was not observed at 6× or 32× MIC up to 96 h. Based on these results, we conclude that MMV688844 is a bacteriostatic agent.
In summary, we screened the Pathogen Box in a statistically confident screening model and identified several hits in the dose-response curve. Among these hits, MMV688844 showed the best in vitro activity against wild-type M. abscessus and various clinical isolates. MMV688844, which was found to have acceptable cytotoxicity, will be used in further studies as a lead compound in the development of new treatment options for pulmonary M. abscessus infection. Additional studies will concentrate on the structure-activity relationship, target deconvolution, and in vivo efficacy.