Turning the respiratory flexibility of Mycobacterium tuberculosis against itself

Abstract

The Mycobacterium tuberculosis (Mtb) electron transport chain (ETC) has received significant attention as a drug target, however its vulnerability may be affected by its flexibility in response to disruption. Here we determine the effect of the ETC inhibitors bedaquiline, Q203 and clofazimine on the Mtb ETC, and the value of the ETC as a drug target, by measuring Mtb's respiration using extracellular flux technology. We find that Mtb's ETC rapidly reroutes around inhibition by these drugs and increases total respiration to maintain ATP levels. Rerouting is possible because Mtb rapidly switches between terminal oxidases, and, unlike eukaryotes, is not susceptible to back pressure. Increased ETC activity potentiates clofazimine's production of reactive oxygen species, causing rapid killing in vitro and in a macrophage model. Our results indicate that combination therapy targeting the ETC can be exploited to enhance killing of Mtb.

Figures

Figure 1. Diagram of the Seahorse XF Analyzer, its function and the initial bioenergetics analysis of Mtb in the presence of the ETC inhibitors.

(a) Compounds are delivered into microplate wells via drug ports. When the probe is lowered, a transient microchamber is formed above a monolayer of bacilli. Dissolved O2 and pH are monitored by sensing probes. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) are calculated from these measurements by the instrument software. (b) ECAR represents carbon catabolism and TCA cycle activity, which produce reducing equivalents that enter the ETC. Reducing equivalents pass through NDH2 or other dehydrogenases (DHs) to the menaquinone pool (MK), and then through Complexes III (cytochrome bc1) and IV (cytochrome aa3), or through cytochrome bd to O2. This contributes to the PMF, which powers ATP synthesis by Complex V (ATP synthase). CFZ acts on NDH2. Q203 inhibits Complex III. BDQ inhibits Complex V. (c) Bioenergetic analysis of Mtb. At the indicated times, 2 g l−1 of glucose (Glc) was added, followed by BDQ, Q203, CFZ, or other drugs, followed by the uncoupler CCCP to stimulate maximum respiration. BDQ and Q203, unlike CFZ or standard antimycobacterial drugs, induce an increase in bacterial respiration, above that of their respective vehicle controls. OCR and ECAR are indicated as a percentage of baseline values. Standard deviation of three replicate wells are indicated as calculated by the Seahorse XF Wave software. One representative experiment is shown; for ETC targeting drugs, at least three replicate experiments were performed. The following inter-experiment % CVs were calculated using Microsoft Excel (Microsoft Office 2010): basal OCR 47.2±5.2; % CV 11.1 (n=6), increased OCR after BDQ addition; 129.7±5.2, % CV 4.1 (n=4) and increased OCR after Q203 addition; 104.1±11.1, % CV 10.7 (n=4). The absolute value OCR profiles are shown in Supplementary Fig. 1. During optimization it was determined that fewer than 5% of the bacteria seeded into microplate wells were dislodged from the bottom during the experiment. (d) O2 depletion from microchamber sustained for 2 h. Each condition was repeated at least four times; representative traces are shown.

Figure 2. IMV, ROS and membrane potential experiments.

ΔpH measured by quenching of fluorescent probe ACMA, ATP production measured with luciferase/luciferin system, the rate NADH consumption was monitored at 340 nm, ROS production via flow cytometry with ROS-sensitive dyes dihydroethidium (DHE), CellROX Orange and CellROX Green, and bacterial membrane potential with the BacLightTM Membrane Potential kit via flow cytometry. (a) ΔpH generated by IMVs in the presence of NADH, with ADP and phosphate in the reaction buffer. (b) ΔpH generated by IMVs in the presence of NADH, without ADP or phosphate. (c) Proton conductance of IMVs. ΔΨ was generated by valinomycin treatment of IMVs in potassium-free buffer, ΔpH generation was taken as an indication of membrane permeability to protons. (d) ATP production by IMVs provided with NADH. (e) Rate of NADH consumption by IMVs in the presence of saturating concentrations of NADH (250 μM). (f) Median fluorescent intensity of ROS-sensitive dyes in treated Mtb mc26230 cells. (h) Shifts in the red/green median fluorescent intensity ratio after control and drug addition shows that neither BDQ nor Q203 have a measurable effect on Mtb membrane potential, whereas CCCP does. Error bars are standard deviations of three replicate experiments, except for subpanel (f), for which the interquartile range of a single experiment is shown, and (g), for which five biological replicates were used. P-values were determined by one-way analysis of variance using GraphPad Prism 6.05. *P<0.05, **P<0.005, ***P<0.0005 and ****P<0.0001.

Figure 3. Bioenergetic analysis of Mtb H37Rv, cytochrome bd knockout (cydKO), and cydKO strain with Q203 resistance SNP (A317V) to ETC drugs.

At indicated times, 2 g l−1 glucose was added, followed by either Q203, BDQ, or DCCD, followed by 2 μM CCCP. (a) 300 × MIC50 BDQ treatment has a very similar effect on H37Rv and cydKO. (b) 300 × MIC50 Q203 treatment causes H37Rv OCR to increase, however cydKO OCR drops to zero, while A317V OCR is unchanged. The untreated media OCR profiles of the Mtb H37Rv, cydKO and A317V strains are provided in Supplementary Fig. 4. (c) 100 μM of the ATP synthase inhibitor DCCD causes a similar OCR increase to 30 × MIC50 BDQ, although somewhat less in magnitude at these concentrations. (d) H37Rv Mtb was cultured for 24 h with 30 × MIC50 of BDQ, Q203, CFZ, a RIF/INH combination, or media control. Relative abundance of ATP, ADP, and AMP was measured by LC-MS/MS. Standard deviations of three technical replicates are shown, all experiments were performed at least twice. *P<0.05, **P<0.005 and ****P<0.0001 (one-way analysis of variance using GraphPad Prism 6.05).

Figure 4. Analysis of combination treatment of Mtb H37Rv.

Indicated additions are 2 g l−1 glucose (Glc), drug combination, and 2 μM CCCP as an uncoupler to stimulate maximum respiration. OCR and ECAR are shown as a per cent of baseline values with standard deviation of three replicate experiments performed. Kill curves were constructed from CFUs measured over 20 days of treated cells grown in 7H9 with 2 g l−1 of glucose and 0.01% tyloxopol. Limit of detection (LOD) is 10 CFU per ml. All drugs were added at 30 × MIC50 values. Two separate kill curve experiments were performed with the same results, representative data is shown. (a) BDQ, Q203, BDQ/Q203, and BDQ/Q203/CFZ combinations. (b) BDQ, CFZ, BDQ/CFZ and BDQ/Q203/CFZ combinations. (c) Q203, CFZ, Q203/CFZ, and BDQ/Q203/CFZ combinations.

Figure 5. Assessment of contributors to rapid synergistic killing of CFZ-containing combinations.

(a) NADH/NAD+ ratios of cells cultured with indicated drug combinations at 30 × MIC50, as measured by LC-MS/MS. (b) ROS produced by cells cultured with indicated drug combinations at 30 × MIC50, as measured by flow cytometry with DHE. (c) The killing efficacy of the BDQ/Q203/CFZ under three different conditions after 5 days of treatment: no antioxidant (NAO), 4-Hydroxy-Tempo (4HT) and N-acetyl-cysteine (NAC). Representative data is shown of three independent experiments performed. Limit of detection (LOD) is 10 CFU per ml. ND, not detected. (d) ATP, ADP, and AMP ratios of cells cultured with indicated drug combinations at 30 × MIC50, as measured by LC-MS/MS. All experiments were performed in triplicate; error bars indicate standard deviation of measurements. *P<0.05, **P<0.005, ***P<0.0005 and ****P<0.0001 (one-way analysis of variance using GraphPad Prism 6.05). Correlation between reduction in CFU per ml (log10) at Day 3 as a measure of early, rapid killing, and (e) reductive stress, (f) oxidative stress, and (g) ATP depletion stress. Correlations are shown for all drugs (dotted line), as well as for ETC-targeting drugs only (solid line). Excluding RIF/INH control, which is expected to have a different mechanism of action, linear correlation statistics were determined using GraphPad Prism 6.05 and are as follows: (e) R2=0.83, P=0.002; (f) R2=0.91, P=0.0002; (g) R2=0.137, P=0.4.

Figure 6. Analysis of the effects of different Mtb ETC targeting drug combinations on mammalian cell lines.

Cell lines were treated with the ETC targeting drugs at 30 × MIC50 concentrations. Oligomycin was used at a concentration of 3.0 μM and 1.5 μM for HepG2 and RAW264.7 cells, respectively. FCCP was used at a concentration of 1.5 μM, and rotenone and antimycin A were both used at a concentration of 0.5 μM. HepG2 cells (a–c) and RAW264.7 cells (d–f) were seeded at a density of 25,000 and 65,000 cells per well, respectively. In both the HepG2 and RAW264.7 cell lines the Mtb ETC targeting drug combination had similar effects on the four bioenergetic parameters as observed in the controls. These parameters include basal respiration (BR), ATP turnover (AT, an indication of the amount of O2 consumed to produce ATP in the mammalian ETC), spare respiratory capacity (SRC, an indication of the cell's capacity to respond under stressful conditions) and non-mitochondrial respiration (non-MR, e.g. NADPH oxidase mediated reduction of O2 during the oxidative burst). Experiments were performed twice. Kill curves (g–i) were constructed by infecting 200,000 RAW264.7 cells per well with Mtb H37Rv at a MOI of 1 and drug treating at 30 × MIC50 concentrations, for 4 days. CFU data is presented as the mean percentage±s.d. reduction in CFU from Day 0 of three independent experiments performed in triplicate.

Figure 7. Proposed models of Mtb energy generation pathways under different drug treatment conditions.

(a) In untreated Mtb the proton motive force (PMF) produced by the terminal oxidases is used by Complex V to produce ATP. AXP ratios allosterically regulate both glycolysis and the TCA cycle via feedback inhibition to maintain energy homeostasis. (b) In the presence of BDQ, Complex V and thus ATP production is inhibited. Although Complex V does not relieve the PMF, PMF is modulated by either intrinsic or extrinsic uncoupling, preventing back pressure. Without OXPHOS, ATP levels decline, relieving carbon catabolism and the TCA cycle of feedback inhibition. This increases production of reducing equivalents, which enter the ETC and increase OCR. As the feedback loop is broken by BDQ, this control system fails to regulate activity appropriately, and glycolysis and the TCA cycle accelerate to their maximum pace. (c) In the presence of Q203, Complex III is inhibited. Electrons reroute from the Complex III/IV pathway to the cytochrome bd pathway, which generates less PMF. This reduces the efficiency of OXPHOS, leading to declining ATP levels, a decrease in feedback inhibition of carbon catabolism and the TCA cycle, increased production of reducing equivalents and increased respiration. The feedback loop is still functional, and a new steady state is reached with reduced ATP levels. (d) In the presence of all three drugs, production of reducing equivalents is increased as discussed above. The increased flux of electrons through NDH2 enhances CFZ's ability to divert electrons to O2, causing the formation of toxic ROS.