Advancing the Therapeutic Potential of Indoleamides for Tuberculosis

Indole-2-carboxamide derivatives are inhibitors of MmpL3, the cell wall-associated mycolic acid transporter of Mycobacterium tuberculosis. In the present study, we characterized indoleamide effects on bacterial cell morphology and reevaluated pharmacokinetics and in vivo efficacy using an optimized oral formulation.

D espite decades of efforts and progress in tuberculosis (TB) control programs and research for new drugs and therapeutic regimens, TB continues to pose a great risk to global public health security as the leading cause of death by a single infectious agent. In 2016, an estimated 10.4 million new cases of TB occurred globally, including 600,000 cases with resistance to rifampin, the most powerful first-line TB drug in use (1). Also in 2016, there were 1.7 million deaths from TB globally, including 0.4 million deaths among people coinfected with HIV (1). Drug resistance has become more prominent because of the high global burden and poor treatment success rate (1,2).
Indoleamides have demonstrated extremely low MICs against both drug-susceptible and drug-resistant M. tuberculosis and activity in mouse infection models (17,18). Indoleamides are bactericidal against M. tuberculosis (17,18), most likely because of damage to the cell wall due to the lack of matured mycolic acids. However, the effect of indoleamide treatment on cell morphology has not been reported. Most interestingly, an expanded-spectrum indoleamide (Fig. 1, compound 2) showed significant synergy with rifampin both in vitro and in a mouse infection model (19), making this chemotype even more promising as a new TB drug candidate. Our objectives for this study were to determine the effect of indoleamide treatment on the cell morphology of M. tuberculosis and improve the pharmacokinetic (PK) profile and in vivo efficacy of the lead compound by optimizing the formulation.

Indoleamide treatment-induced morphological changes of M. tuberculosis.
To determine whether indoleamide treatment affects cell morphology, we treated midlog-phase M. tuberculosis H37Rv cells with 1ϫ and 10ϫ MIC of compound 1 (18) and used untreated cells and cells treated with 10ϫ MIC of isoniazid as controls. After 24 hours of treatment, cells were processed and subjected to field emission scanning electron microscopy (FESEM). Cells treated with compound 1 showed a significantly higher number of unhealthy cells, as exemplified by more dimples at the poles and septum and a wrinkled cell surface near the poles ( Fig. 2A to D). Treatment with compound 1 at 10ϫ MIC resulted in a higher rate of deformed bacilli than 1ϫ MIC, suggesting a dose-dependent effect (Fig. 2E). Treatment with isoniazid (INH) at 10ϫ MIC resulted in more dimples per 100 cells than no treatment, but the difference did not reach statistical significance (Fig. 2E).
An optimized formulation enhanced oral bioavailability. Pharmacokinetic studies in Swiss Webster mice were performed with compound 2 dosed by oral and intravenous (i.v.) injection. An oral propylene glycol (PG):Tween 80 formulation was compared with a carboxymethyl cellulose (CMC) formulation at the 10 mg/kg of body weight dosing level. The optimized PG:Tween 80 formulation significantly improved the pharmacokinetic parameters compared with the universal 0.5% (wt/vol) CMC formulation, as evidenced by improved maximum concentration of drug in serum (C max ) (1.04 versus 0.15 g/ml) and area under the curve to the last measurable concentration (AUC last ) (6.56 versus 1.51 h·g/ml) ( Table 1). The optimized formulation provided a relative bioavailability of 96%, which represents a 74% increase compared with the CMC formulation (22%) ( Table 1). Other formulations, such as 0.5% wt/vol methylcellulose nano-suspension and solid dispersion, were also tested for comparison, but all were inferior to the PG:Tween 80 formulation (see Table S1 in the supplemental material). The plasma concentration of free compound 2 exceeded the MIC (0.004 g/ml) for 24 h postdose with both the oral (p.o.) and i.v. dosing regimens (see Fig. S1 in the supplemental material). The plasma protein binding of compound 2 in mouse was 96.8%.
Improved in vivo efficacy of the optimized formulation in a mouse infection model. Compound 2 was shown to be active in vivo in our previous study (19). We hypothesized that compound 2 prepared in the optimized formulation of PG:Tween 80 (4:1, vol/vol) would have superior activity. Indeed, substantial dose-dependent bactericidal activity was demonstrated by lung CFU counts (Table 2), reduced granuloma  formation (Fig. 3A), and lung weight (Fig. 3B). At 50 mg/kg, 4 weeks of treatment with compound 2 achieved a 4.66-log lower mean CFU count than that of untreated controls and 1.2-log lower CFU count than that of the positive control ethambutol at 100 mg/kg. These findings represent a significant improvement compared with the previous study using the 0.5% (wt/vol) CMC formulation, in which 4 weeks of treatment with 100 mg/kg of compound 2 resulted in CFU counts that were 2.0 log higher than the same positive control (19).
Superior in vivo efficacy of compound 2 compared with compound 3. To compare the in vivo efficacy of compound 2 with that of the published indoleamide analog compound 3 (17), we evaluated their dose-ranging activity in a mouse highdose aerosol infection model. Treatment commenced 2 weeks after implantation of 3.72 log 10 CFU of M. tuberculosis H37Rv, when the mean lung CFU count was 6.27 log 10 . Both compounds showed dose-dependent activity (Fig. 4), with bactericidal effects observed at doses of Ն30 mg/kg (P Ͻ 0.01) (see Table S2 in the supplemental material). At 10 mg/kg, compound 2 significantly reduced the lung CFU burden compared with untreated controls (P Ͻ 0.05), but compound 3 did not. On comparison of the sigmoidal dose-response curves, the 50% effective dose (ED 50 ) was statistically significantly lower for compound 2 (28.9; 95% confidence interval [CI], 25.0 to 33.4) than for compound 3 (43.0; 95% CI, 36.9 to 50.1) (P Ͻ 0.0001), indicating the superior in vivo potency of compound 2 ( Fig. 4; Table S2).

DISCUSSION
MmpL3 is an emerging target for new antituberculosis drugs. Indoleamide inhibitors of MmpL3 previously demonstrated in vitro and in vivo activity warranting further development of this scaffold (10,11,(17)(18)(19). In this study, we observed the effect of indoleamide treatment on M. tuberculosis cell morphology in vitro. Dimples on the poles and septum and a wrinkled cell surface near the poles were induced in a dose-dependent manner, consistent with the findings by Carel et al. in a mycobacterial model system that MmpL3 specifically and dynamically accumulated at the poles and septa during bacterial growth (20). Moreover, the dynamic colocalization of MmpL3 with Wag31, the protein responsible for polar localization of mycobacterial peptidogly-   can biosynthesis, suggests that the main components of the mycomembrane may potentially be synthesized at these precise loci (20). Wag31 proteomic interactome pulldown analysis identified MmpL3, among five other proteins, as a binding partner, and implicated MmpL3 in cell wall lipid permeability and multiple lipophilic antibiotic resistance (21). These characteristics justify MmpL3 as a potential therapeutic target for novel inhibitors, such as indoleamides.
We optimized the formulation for oral administration and reevaluated a representative expanded-spectrum indoleamide, compound 2, achieving significant improvements in bioavailability and in vivo efficacy. Compound 2 is insoluble in water with a predicted partition coefficient (clogP) of 5.74. In our previous study using the universal 0.5% (wt/vol) CMC formulation, compound 2 showed therapeutic efficacy in a mouse infection model, but dosing at 100 mg/kg did not achieve efficacy comparable to the ethambutol control (19). In the current study, a 50-mg/kg dose in the optimized formulation achieved far superior in vivo efficacy compared with the same ethambutol control, demonstrating the in vivo bactericidal potential of this class. Because of the suboptimal in vivo activity of compound 2 in previous work (19), we used bacteriostatic ethambutol rather than bactericidal INH as a positive control in the current study. Although monotherapy with 50 mg/kg of compound 2 did not lower the lung CFU burdens in the day 1 model, bactericidal activity (log CFU reduction) was achieved in the day 14 acute model for both compound 2 and compound 3. This could be attributable to higher dose levels (100 and 200 mg/kg) and the impact of the host adaptive immune response.
Two independent groups have developed and characterized indoleamides as a scaffold targeting MmpL3 of M. tuberculosis in vitro and in vivo simultaneously (10,11,17,18). In this study, we compared two expanded-spectrum indoleamide lead compounds from these two programs, compound 3 and compound 2, side-by-side in a mouse infection model. While both compounds exhibited bactericidal activity at doses of 30 mg/kg and above, compound 2 showed statistically significantly greater potency (i.e., lower ED 50 ) than compound 3, suggesting that, if similar exposures of the two compounds could be safely attained in humans, then compound 2 should be more active.
Indoleamides are active not only on drug-susceptible but also on drug-resistant M. tuberculosis (17,18). Most interestingly, indoleamides showed synergy with the first-line TB drug rifampin in a mouse infection model (19). In addition, the spontaneous resistance frequency is rather low (17,18). The improved efficacy demonstrated with the optimized formulation improvement in this study makes this structural class even more promising for further development as a novel component in new regimens to treat drug-susceptible and drug-resistant TB.
Field emission scanning electron microscopy. To study the effect of indoleamide on mycobacterial cell morphology, M. tuberculosis H37Rv was cultured to optical density at 600 (OD 600 ) of 0.06. Compound 1 (Fig. 1) (18) was added to a final concentration of 0.004 (1ϫ MIC) or 0.04 g/ml (10ϫ MIC). Isoniazid (0.4 g/ml) was used as the control. Samples were collected after 24 h of incubation. Bacteria were immobilized to poly-L-lysine-charged coverslips for 30 min and processed for field emission scanning electron microscopy (FESEM), as described previously (22). For quantification, 100 cells from at least 10 fields from each of three biological replicates were counted under a fixed magnification.
Pharmacokinetic study. To enhance oral bioavailability, we developed a novel formulation for compound 2 in propylene glycol (PG):Tween 80 (4:1, vol/vol). We compared the bioavailability of this new formulation with a previously reported formulation using 0.5% (wt/vol) carboxymethyl cellulose (CMC) (19). For oral (p.o.) administration, compound 2 was prepared in either formulation at 1 mg/ml of homogeneous suspension to achieve a 10-mg/kg dose level. In vivo efficacy evaluation. Four-to 6-week-old female BALB/c mice were obtained from Charles River Laboratories. Mice were aerosol infected with M. tuberculosis H37Rv using an inhalation system (Glas-Col Inc., Terre Haute, IN). At day 1 and day 3 postinfection, 3 mice were sacrificed to determine the number of CFU implanted and the pretreatment CFU burden in the lungs, respectively. From day 3 after infection, groups of 5 mice were treated with 6.25, 12.5, 25, or 50 mg/kg of compound 2 by daily gavage (5 days per week). Ethambutol was administered at 100 and 50 mg/kg to positive controls. Infected but untreated mice were used as negative controls. Compound 2 was prepared with the improved PG:Tween 80 (4:1 vol/vol) vehicle. At days 14 and 28 after treatment initiation, 5 mice from each treatment group were sacrificed and the lungs removed. At the latter time point, lungs were photographed for gross pathology comparison. Lungs were homogenized, diluted, and plated onto 7H11 selective agar plates to enumerate CFUs.
To compare the in vivo efficacy and dose-response relationships of compound 2 and compound 3 ( Fig. 1) (17), female BALB/c mice, 5 weeks of age, were aerosol infected with approximately 5,000 CFUs of M. tuberculosis H37Rv. Treatment started 2 weeks after infection. Both compounds were prepared in a lipid-based microemulsion preconcentrate (MEPC) formulation previously developed for NITD-304 (17). Doses ranged from 3, 10, 30, 100, to 200 mg/kg. Mice were dosed daily gavage (5 days/week) for 4 weeks. Five mice per treatment group were sacrificed after 4 weeks of treatment for lung CFU enumeration.
The Institutional Animal Care and Use Committee of the Johns Hopkins University School of Medicine approved all animal procedures performed in this study.
Statistical analysis. Lung CFU counts (x) were log transformed (as x ϩ 1) before analysis. Group mean CFU counts were compared using one-way analysis of variance (ANOVA) with Dunnett's posttest to adjust for multiple comparisons. Dose-response curves were fit by nonlinear regression using a four-parameter inhibitory sigmoidal maximum effect (E max ) equation, assuming the same E max for both compounds. ED 50 values were compared using the extra sum-of-squares F test. All analyses were performed with GraphPad Prism version 6 (GraphPad, San Diego, CA).

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AAC .00343-19.

ACKNOWLEDGMENTS
We thank Nader Fotouhi of TB Alliance for his critical discussion. This work was supported by TB Alliance with support from Australia's Department of Foreign Affairs and Trade, the Bill & Melinda Gates Foundation (OPP1129600), Irish Aid, Netherlands Ministry of Foreign Affairs, United Kingdom Department for International Development, and United Kingdom Department of Health. NIH funding grants AI37856 and HL133190 to W.R.B. supported this work, along with additional support from the TB Alliance to E.L.N.