Dual inhibition of the terminal oxidases eradicates antibiotic‐tolerant Mycobacterium tuberculosis

Abstract The approval of bedaquiline has placed energy metabolism in the limelight as an attractive target space for tuberculosis antibiotic development. While bedaquiline inhibits the mycobacterial F1F0 ATP synthase, small molecules targeting other components of the oxidative phosphorylation pathway have been identified. Of particular interest is Telacebec (Q203), a phase 2 drug candidate inhibitor of the cytochrome bcc:aa3 terminal oxidase. A functional redundancy between the cytochrome bcc:aa3 and the cytochrome bd oxidase protects M. tuberculosis from Q203‐induced death, highlighting the attractiveness of the bd‐type terminal oxidase for drug development. Here, we employed a facile whole‐cell screen approach to identify the cytochrome bd inhibitor ND‐011992. Although ND‐011992 is ineffective on its own, it inhibits respiration and ATP homeostasis in combination with Q203. The drug combination was bactericidal against replicating and antibiotic‐tolerant, non‐replicating mycobacteria, and increased efficacy relative to that of a single drug in a mouse model. These findings suggest that a cytochrome bd oxidase inhibitor will add value to a drug combination targeting oxidative phosphorylation for tuberculosis treatment.

2. The author should explain why treating with Q203 alone could enhance OCR compared to control in figure 1d and 1e, however in figure 1f, the OCR is reduced significant by Q203 compared to control? In figure S2b-e, there seems no difference between Q203 treatment and control.
3. In Page 8, the authors state that "escape mutants resistant to ND-011992 could not be isolated". This is lack of sufficient support. Though they found 18 escape mutants selected on ND-011992+Q203 are resistant to Q203, this cannot rule out the situation that they are resistant to both inhibitors. The authors should provide data like figure S5 by testing with ND-011992 alone.
4. Page 6 line 6 from bottom: how is 110-fold calculated? The data in figure 2c and Table S2 seem not match with this result. 12. Figure 2a: the gene names are not readable. please enlarge this panel.
13. Figure 3: the order of panel should be adjusted to follow the citation order in the maintext. In the legend, line 7, Q203 is missing after 100 nm. You should keep one decimal space after the decimal point for all the data in panel d.
14. Figure S2c,2e: mark the significant of differences through statistic test. 15. Table S1: SD value is missing. 16. Table S2: is the value of 3.54 correct?
Referee #2 (Remarks for Author): In their paper entitled "Dual inhibition of the terminal oxidases eradicates antibiotic-tolerant Mycobacterium tuberculosis", Lee et al. report on the use of a whole-cell screen approach to identify a small molecule inhibitor of cytochrome bd (ND-011992), which, when used in combination with telacebec (Q203), an inhibitor of the cytochrome bcc:aa3 terminal oxidase, shows promising activity against replicating mycobacteria and antibiotic-tolerant persisters in various in vitro stress models and in a mouse model of acute TB. The paper addresses a very important area of TB pathogenesis and the need for developing novel strategies to target Mtb persisters, which are believed to account for the difficulty in eradicating infection with the current standard-of-care regimen. Strengths of the study include the focus on oxidative phosphorylation, which appears to be a critical pathway for Mtb viability under both replicating and nonreplicating conditions, as well as the extensive studies undertaken to characterize the on-target effects of the compound, including studies with deletion and overexpression strains and transcriptomics analyses. Also, the activity of ND-011992 was tested against various clinical isolates, including drug-susceptible and drugresistant isolates. Several issues outlined below diminish the overall enthusiasm for the paper in its current form.
Major comments: 1. The Introduction and conclusions focus on Mtb, but various experiments have been performed in three different species (M. tuberculosis, M. bovis BCG, or M. smegmatis) without a clear justification in the main text, or even a clear indication of which strain is being used (even in the figure legends it is sometimes unclear, e.g., Fig. 1e). It is not clear, for example, why the inverted membrane vesicle studies were performed in M. smegmatis (Supplementary Figure 4). Also, the authors do not address why there is a significant discrepancy in the IC50 of ND-011992 required to deplete ATP levels in combination with Q203 in BCG vs. Mtb (Figure 1b and Supplementary Figure 1a), as well as the minimum FIC index in the two species ( Supplementary Figures 1b and 1c). Does the compound show higher affinity for Cyt-bd from BCG vs. Mtb? What is the sequence homology and structural similarity between the two genes/proteins? 2. Based on experience with other chemical library screens, it is highly remarkable that ND-011992 was identified from a collection of 53 small molecules. How were these compounds selected? Were they in any way enriched for potential hits? 3. For the methylene blue assays, it would be helpful to have a media-only control to understand the background signal. 4. Fig 1 d-f: Indicate to what is %OCR normalized. 5. Fig 1f : This would be more convincing if the cydAB gene deleted and complemented were from the same species. Also, an H37Rv ΔcydAB strain is later used (Fig 3b), for which complementation data are lacking. 6. Fig 2a: Which genes are significantly regulated? This information could be included in the supplement. Also, it is stated that the transcriptional response of H37Rv treated with the Q203/ND-011992 combination was almost identical to the response of H37Rv ΔcydAB treated with Q203 alone, but it is not clear what % of genes are actually overlapping between these two groups. Also, what are the pathways represented by the ~200 genes that are differentially regulated in the mutant treated with Q203 alone vs. the WT treated with the two drugs? 7. Fig 4a-c: A Q203-alone control is not shown, so it is difficult to determine whether the activity of the combination is truly synergistic, as described, or simply additive (ND-011992 has very little activity against actively replicating Mtb, as well as against nutrient-starved and hypoxic Mtb). Judging by the in vivo data (Fig 4d), the combination appears to be additive, but this can be formally calculated. 8. In vivo studies: The argument is made that pharmacological inhibition of both the cytochrome bcc:aa3 terminal oxidase and the cytochrome bd oxidase is a favorable strategy to target antibiotic-tolerant Mtb in order to shorten treatment length, however the animal model used is a very short-term model (5 days), in which the bacteria are actively replicating (and not antibiotictolerant). In addition, the study uses intranasal infection and C56BL/6 mice, which develop highly cellular lesions. A more appropriate model to truly evaluate the activity of this compound combination on antibiotic-tolerant Mtb would be a chronic infection model (initiation of treatment at least 4-6 weeks after (preferably aerosol) infection). Also, the C3HeB/FeJ mouse model would be preferable since, unlike C57BL/6 mice, it develops human-like necrotic granulomas, which, because of tissue hypoxia and other stress conditions, are believed to promote antibiotic tolerance in the extracellular bacteria found in the caseous core. Also, this model would offer a way of testing the penetration of these compounds into pathological lesions of clinical relevance.

Summary and General Comments
The manuscript "Dual inhibition of the terminal oxidases eradicates antibiotic-tolerant Mycobacterium tuberculosis" is an exciting study describing the identification of a cytochrome bd inhibitor which synergises with Q203 and they quite convincingly shows targets cytochrome bd oxidase and presents an excellent approach to targeting oxidative phosphorylation for tuberculosis treatment. The drug combination showed good activity in vitro and also in a murine model. I have some minor comments which are discussed below. Minor points (1) In the abstract the authors state "The drug combination was bactericidal against replicating mycobacteria and antibiotic tolerant Persisters" however the experimental data shows that the drugs work on populations of cells in models which reduce the growth rate of Mtb. Whilst these experiments show the efficacy of the drugs against POPULAQTIONS of cells which are tolerant to other anti-TB drugs the experiments do not measure killing of the specialised sub-population of persister cells. To do this the authors would need to perform kill curves to test for changes in the biphasic kill curve (see more about this below). This just need rephrasing so that it doesn't mislead the reader. This also needs to be considered throughout the rest of the document.
(2) It is interesting that the drug combination was more effective against BCG as compared with Mtb. Why do the authors think this is the case? It seems worthy of a line of discussion. And considering this its strange that for the identified resistant strain the authors only investigated BCG. Were the Mtb drug resistant strains also investigated?
(3) For some of the experiments there needs to be some justification on the strain use as data seems to be presented from BCG, Mtb and M. smegmatis in Figure 1.
(4) For the experiments where the authors measure the killing efficiency of the drug combinations in the Betts starvation model and a hypoxic model the authors pick one time point. Although there is an inoculum bar there doesn't seem to be a drug free control on figure 4b and 4c whereas there is a control on the growing cells (4a) and there is no Izoniazid control on figure 4a which are important and should be added. Im convinced that the drug combination works on these slow growing drug tolerant cells but without a time course the authors cannot comment on the effect on the persister population.
1. The authors proposed a way of dual inhibition of the terminal oxidases by the ND-011992-Q203 combination for more effective treatment of TB. Since there is already an approved drug bedaquiline which also targets the oxidative phosphorylation pathway. The author should have more analysis or discussion on comparison of the two drug-therapy ways. For example, what are the differences on the performance between ND-011992-Q203 and bedaquiline? Is there any potential advantage or complementary effect of ND-011992-Q203 combination on bedaquiline? Can we use ND-011992-Q203-bedaquiline three drug combination to enhance anti-TB efficacy?
We agree with this comment: it will be important to evaluate if bedaquiline (or other approved drugs) could further enhance the bactericidal potency of the Q203/ND-011992 combination. The following paragraph was added in the discussion section (Lines 340-345): "Future studies are needed to identify drugs capable of enhancing further the bactericidal potency of the combination Q203/ND-011992. Since our results indicate that the combination Q203/ND-011992 is at least as bactericidal as bedaquiline against replicating and nonreplicating M. tuberculosis, it will be particularly interesting to test the bactericidal potency of the 3-drug combination first in vitro, and in animal models after the development of an improved version of ND-011992." 2. The author should explain why treating with Q203 alone could enhance OCR compared to control in figure 1d and 1e, however in figure 1f, the OCR is reduced significant by Q203 compared to control? In figure S2b-e, there seems no difference between Q203 treatment and control.
The experimental setting varied significantly between the experiments represented on those figure panels. In Figure 1D-E, the OCR measurements were performed on a Seahorse XFe96 Analyzer. The result reflects rapid changes immediately after the addition of Q203. The mechanisms that drive this initial spike in OCR detected with the Seahorse Xfe96 analyser is still unclear but has already been reported by some of us (Lamprecht et al, 2016). It likely reflects an immediate stress response upon chemical inhibition of the Cyt-bcc-aa 3 characterized by a transient increase in activity of the Cyt-bd. Interestingly, this phenomenon is also observed upon treatment with bedaquiline (Lamprecht et al., 2016).
In the Figure EV2B-E (previously Supplementary figure 2b-e), the OCR was quantified using the MitoXpress® Probe based on readings collected over a few hours, and represent a cumulative, averaged OCR. The assay also included a period of 10 min preincubation before OCR is recorded. While the Seahorse XFe96 analyzer operates at constant oxygen concentration (the microchamber is mixed every few minutes to maintain high oxygen concentration (see Methods and Materials section), the MitoXpress assay does not. These differences in methodology may explain why an initial spike in OCR is not observed using the MitoXpress assay. Figure 1F involved testing in inverted membrane vesicles (IMVs) from Mycobacterium smegmatis. The use of IMVs may enhance the potency of Q203 as the target QcrB is more readily accessible to chemical inhibition by Q203 and hence inhibition of OCR is the 6th Oct 2020 1st Authors' Response to Reviewers dominant effect observed in these experiments. The following sentence was added to the main text (Lines 129-131): "The use of IMVs may enhance the potency of Q203 as the target QcrB is more readily accessible to chemical inhibition by Q203 and hence amplifying the effect of OCR inhibition in these experiments." 3. In Page 8, the authors state that "escape mutants resistant to ND-011992 could not be isolated". This is lack of sufficient support. Though they found 18 escape mutants selected on ND-011992+Q203 are resistant to Q203, this cannot rule out the situation that they are resistant to both inhibitors. The authors should provide data like figure S5 by testing with ND-011992 alone.
We initially assumed that the frequency of co-resistance was too low to think that the escape mutants resistant to Q203 could be resistant to ND-011992 as well. To address this concern, we conducted an additional experiment. It was previously reported that lansoprazole, a Cytbcc:aa 3 inhibitor, remains potent against Q203-resistant mycobacteria (Rybniker et al, 2015). After confirming that lansoprazole was potent against the 18 mutants selected on Q203 + ND-011992 (Table EV1), we tested the ability of ND-011992 to deplete intracellular ATP in the 18 escape mutant strains in the presence of a fixed dose of lansoprazole. Results confirmed that ND-011992 was still active in all isolated mutant strains, indicating that the escape mutants are susceptible to ND-011992 (Table EV1).
The following sentences were added to the main text (Lines 214-225): "All 18 M. bovis BCG colonies analyzed were highly resistant to Q203 (MIC 50 > 100 nM) ( Table EV1), suggesting that the primary mechanism of resistance to the drug combination is mediated by a mutation in QcrB (Pethe et al., 2013). To exclude that any of the escape mutants were co-resistant to ND-011992, we made use of lansoprazole, an inhibitor of the mycobacterial cytochrome bcc:aa 3 that retains efficacy against Q203-resistant spontaneous mutants (Rybniker et al., 2015). Upon confirmation that lansoprazole retained efficacy against the spontaneous resistant mutants (Table EV1), we tested the inhibitory activity of a dose-range of ND-011992 in the same intracellular ATP assay used in the initial screen ( Figure 1B), but in the presence of lansoprazole instead of Q203. ND-011992 depleted intracellular ATP levels at IC 50 values comparable to those observed in the parental M. bovis BCG strain (Table EV1), indicating that the escape mutants remained fully susceptible to ND-011992." 4. Page 6 line 6 from bottom: how is 110-fold calculated? The data in figure 2c and Table S2 seem not match with this result.
The 110-fold increase was calculated using the data obtained in one of the experimental repeats. We amended to include a range of values calculated from experimental repeats (Lines 166-169): "Consistently, ND-011992 inhibited the growth of an H37Rv strain deficient in Cyt-bcc:aa 3 expression (Beites et al, 2019) at a concentration 33-to 110-fold lower compared to the concentration required to inhibit the growth of the parental strain ( Figure 2C and Appendix Table S2)." 5. Page 6 line 12-13 from bottom: these data are not consistant with Figure 2b.
The following correction has been made to reflect the correct values (Lines 160-162): "Increasing the gene copy number of the cydABDC operon from a high-copy number plasmid caused an 18-to 31-fold shift in ND-011992 potency from 0.52 to 17.2 μM, as well as a 12fold shift in OCR from 0.62 to 7.8 μM ( Figure 2B)." 6. Page 7 line 7 from top: a typo? 523 nm or 532 nm?
The typographical error was corrected in the following sentence (Lines 177-179): "Similarly, NADH-driven reduction of the cytochrome bcc:aa 3 oxidase hemes (a at 443 and 600 nm, b at 432 and 563 nm, and c at 523 and 552 nm) was also clearly assigned for the parental M. smegmatis mc 2 155 IMVs ( Figure EV3A)." 7. Page 7 line 9 from bottom: from {less than or equal to} 0.2 to 2 µM The typographical error was corrected in the following sentence (Line 199): "ND-011992 was potent at an MIC value ranging from ≤ 0.2 to 1 μM against pan-susceptible clinical isolates…" 8. Page 12: the compound name ND-011992 should be consistent throughout the paper.
The schematic diagram under the same subsection was also updated to include the corrected compound name. 9. Page 14-15: use g rather than RPM.
The centrifuge speeds were changed to the unit of × g in the following sentences: Lines 527-528: "Samples were then centrifuged at 16,200 × g

for 15 min at 4 C and 400 μL of the top aqueous layer was carefully collected."
Lines 530-531: "The supernatant was pipetted off and discarded and the pellet was washed three times with ice cold 70% EtOH, and centrifuged at 16,200 × g for 5 min."  13. Figure 3: the order of panel should be adjusted to follow the citation order in the maintext. In the legend, line 7, Q203 is missing after 100 nm. You should keep one decimal space after the decimal point for all the data in panel d.
The panels in Figure 3 was reordered to reflect their citation order in the main text. 14. Figure S2c,2e: mark the significant of differences through statistic test. Figures EV2C and EV2E (previously S2c and S2e) and their corresponding legends were updated to include the significance of differences.
Appendix Table S1 was amended to include the standard deviation of the measurements. 16. Table S2: is the value of 3.54 correct?
Yes, the value is correct, complementation appeared to be partial in this set of experiment.
Referee #2 (Remarks for Author): Several issues outlined below diminish the overall enthusiasm for the paper in its current form.
Major comments: 1. The Introduction and conclusions focus on Mtb, but various experiments have been performed in three different species (M. tuberculosis, M. bovis BCG, or M. smegmatis) without a clear justification in the main text, or even a clear indication of which strain is being used (even in the figure legends it is sometimes unclear, e.g., Fig. 1e). It is not clear, for example, why the inverted membrane vesicle studies were performed in M. smegmatis (Supplementary Figure 4). Also, the authors do not address why there is a significant discrepancy in the IC50 of ND-011992 required to deplete ATP levels in combination with Q203 in BCG vs. Mtb (Figure 1b and Supplementary Figure 1a), as well as the minimum FIC index in the two species ( Supplementary Figures 1b and 1c). Does the compound show higher affinity for Cyt-bd from BCG vs. Mtb? What is the sequence homology and structural similarity between the two genes/proteins?
We agree that the rationale behind the use of different mycobacterial species should have been better explained throughout the manuscript.
For practical biosafety reasons, the assay was initially developed in M. bovis BCG. M. bovis BCG is an excellent surrogate mycobacterial species to study the terminal oxidases since its Cyt-bcc:aa 3 and Cyt-bd share 100% sequence similarity with the M. tuberculosis H37Rv counterparts (Brosch et al, 2007;Lew et  For added clarity, the species used for each experiment was mentioned in the text and in the figure legends. For consistency, the results presented in Figure 1C (obtained with M. tuberculosis H37Rv) was moved to Figure EV2A and substituted with the M. bovis BCG results (shown in Supplementary Figure 2a in the initial submission).
The reason why Mycobacterium smegmatis was used in the IMV experiments is because the model is well-established for assessing biochemical activities of the entire respiratory chain in mycobacteria, and to provide confirmation of target engagement in a membrane-only environment (Hards et al, 2015;Lu et al, 2015;Pecsi et al, 2014). Given that the M. tuberculosis Cyt-bd has high sequence identity with the M. smegmatis homologue (78% and 68% for CydA and CydB, respectively) we consider this biochemical system to be meaningful. The data obtained with M. smegmatis IMVs was in agreement with whole cell data for M. tuberculosis and M. bovis BCG.
To clarify the rationale behind the use of multiple mycobacteria species in the study, we have added the following sentences to the main text: Lines 88-92: "For practical biosafety reasons, the assay was initially developed in M. bovis BCG. M. bovis BCG is an excellent surrogate mycobacteria to study the terminal oxidases since its Cyt-bcc:aa 3 and Cyt-bd share 100% sequence similarity with the M. tuberculosis H37Rv counterparts (Brosch et al., 2007;Lew et al., 2011;Data ref: Garnier T., 2006;Data ref: Lew J. M., 2012)." Lines 119-124: "Biochemical assays were performed on purified inverted-membrane vesicles (IMVs) from Mycobacterium smegmatis mc 2 155 to obtain preliminary evidence that ND-011992 acts on the Cyt-bd in the lipid-rich environment of the IMVs. IMVs purified from M. smegmatis were used since protocols for their purification in high amount and subsequent characterization of the associated electron-transport chain are well-established (Hards et al., 2015;Heikal et al, 2016;Lu et al., 2015;Pecsi et al., 2014)." Figure 1E was corrected to the following to clarify the strain used in the experiment:

"(E) Dose-dependent inhibition of M. bovis BCG OCR by ND-011992 (in the presence of 100 nM Q203) measured on a Seahorse XFe96 analyser platform."
2. Based on experience with other chemical library screens, it is highly remarkable that ND-011992 was identified from a collection of 53 small molecules. How were these compounds selected? Were they in any way enriched for potential hits?
The compounds were selected from a long-standing NIH project that was carried out in the laboratory of Prof. Marvin Miller. Prof. Marvin Miller discovered several chemical series targeting QcrB (Moraski et al, 2012;Moraski et al, 2014;Tiwari et al, 2013). These 53 compounds were previously screened against M. tuberculosis H37Rv (MABA assay) and were selected based on their lack of potency, which is what we expected from a Cyt-bd inhibitor.
"The small-molecules were selected from a larger library assembled during a NIH project focused on the design and discovery of novel small-molecules for tuberculosis treatment (Moraski et al., 2012;Moraski et al., 2014;Tiwari et al., 2013)." 3. For the methylene blue assays, it would be helpful to have a media-only control to understand the background signal.
As described in the manuscript, this assay is not quantitative but has the advantage to provide a rapid visual indication of the combined effect of Q203 + ND-011992 on oxygen consumption. As such, we have not added a medium-only control in the initial experiments. Instead, we relied on the Seahorse XFe96 analyser and the MitoXpress® probe to provide quantitative value of the extent of OCR inhibition by Q203 + ND-011992.
Nevertheless, we addressed this comment by performing a new experiment including a medium-only control in M. bovis BCG shown as Figure 1C in the revised manuscript. The same experiment could not be performed on M. tuberculosis because we do not have access to a BSL-3 laboratory since the beginning of the COVID-19 pandemic.
4. Fig 1 d-f: Indicate to what is %OCR normalized.
The figure legends of Figure 1D-F was edited to include the information:  Fig 1f: This would be more convincing if the cydAB gene deleted and complemented were from the same species. Also, an H37Rv ΔcydAB strain is later used (Fig 3b), for which complementation data are lacking.

"(D) Oxygen consumption assay in M. bovis BCG using the Seahorse XFe96 Extracellular flux analyser. 12 μM ND-011992 was injected alone (green triangles), 100 nM Q203 (blue circles), or in combination with Q203 (red squares
We deemed the results convincing as they demonstrate that ND-011992 can indeed inhibit the M. tuberculosis Cyt-bd in the lipid-rich environment of the IMVs. This experiment was performed with all the appropriate controls providing the conclusive proof. The following sentence was added in the main text (Lines 124-127):

"To provide further evidence of target engagement of ND-011992 with the M. tuberculosis Cyt-bd in a lipid-rich environment, we prepared IMVs from M. smegmatis ΔcydAB mutant expressing the M. tuberculosis cydABDC operon."
In Figure 3D (previously Figure 3b), we used the H37Rv ΔcydAB strain as a reference to compare the Fluctuation analysis of Q203 + ND-011992. Ideally, the H37Rv parental strain should have been used. However, it could not be done since, under the experimental conditions used to conduct the fluctuation analysis (at least 25 days of incubation at 37ºC), a lawn would have been observed using a parental or cydAB-complemented strain due to the bacteriostatic nature of Q203 and loss of potency on glycerol-supplemented agar plates (Kalia et al, 2019). We added the following sentence (Lines 228-233): "In this experiment, we used the H37Rv ΔcydAB strain to compare the frequency of resistance to Q203 + ND-011992 to a reference. The H37Rv parental strain could not be used due to the bacteriostatic nature of Q203 and the loss of potency of the drug on agar plates supplemented with glycerol (Kalia et al., 2019). In addition, rifampicin was added as an additional control to compare the frequency of resistance to Q203 + ND-011992 to an approved drug." 6. Fig 2a: Which genes are significantly regulated? This information could be included in the supplement. Also, it is stated that the transcriptional response of H37Rv treated with the Q203/ND-011992 combination was almost identical to the response of H37Rv ΔcydAB treated with Q203 alone, but it is not clear what % of genes are actually overlapping between these two groups. Also, what are the pathways represented by the ~200 genes that are differentially regulated in the mutant treated with Q203 alone vs. the WT treated with the two drugs?
In order to allow readers to determine which genes are differentially regulated in response to drug challenge, we have included a searchable Microsoft Excel document containing all of the processed differential gene expression data (Source data to Figure 2A).
To address questions about degree of overlap we have included the sentence (Lines 150-153): "The majority of these genes were downregulated, and of the 891 downregulated genes in the combination-treated H37Rv strain, more than two-thirds (67.6%) were similarly downregulated in the ∆cydAB mutant challenged with Q203 alone." The extra 200 genes differentially regulated in the Q203-treated-∆cydAB compared to combination-treated-H37Rv encompassed a wide diversity of pathways without any clear representation worthy of discussion here.
7. Fig 4a-c: A Q203-alone control is not shown, so it is difficult to determine whether the activity of the combination is truly synergistic, as described, or simply additive (ND-011992 has very little activity against actively replicating Mtb, as well as against nutrient-starved and hypoxic Mtb). Judging by the in vivo data (Fig 4d), the combination appears to be additive, but this can be formally calculated.
In Figure 4A-C, the Q203-alone control is shown in blue inverted triangles. 8. In vivo studies: The argument is made that pharmacological inhibition of both the cytochrome bcc:aa3 terminal oxidase and the cytochrome bd oxidase is a favorable strategy to target antibiotic-tolerant Mtb in order to shorten treatment length, however the animal model used is a very short-term model (5 days), in which the bacteria are actively replicating (and not antibiotic-tolerant). In addition, the study uses intranasal infection and C56BL/6 mice, which develop highly cellular lesions. A more appropriate model to truly evaluate the activity of this compound combination on antibiotic-tolerant Mtb would be a chronic infection model (initiation of treatment at least 4-6 weeks after (preferably aerosol) infection). Also, the C3HeB/FeJ mouse model would be preferable since, unlike C57BL/6 mice, it develops human-like necrotic granulomas, which, because of tissue hypoxia and other stress conditions, are believed to promote antibiotic tolerance in the extracellular bacteria found in the caseous core. Also, this model would offer a way of testing the penetration of these compounds into pathological lesions of clinical relevance.
We agree with the reviewer that additional models are required to test the potency of the combination Q203 + a cytochrome-bd inhibitor against persisters in an animal model.
We did not do it because, as explained in the manuscript, the pharmacokinetic properties of ND-011992 are sub-optimum. Many parameters such as potency, penetration in eukaryotic cells, volume of distribution, etc. need to be optimized to obtain a drug candidate compatible for testing in more complex in vivo models.
In the context of this article, we were pleased to observe some degree of potency of ND-011992 (in combination with Q203) in an acute mouse model of tuberculosis. As written in the text, chemical optimization is required to develop an advanced drug candidate.
Referee #3 (Remarks for Author): I have some minor comments which are discussed below. Minor points (1) In the abstract the authors state "The drug combination was bactericidal against replicating mycobacteria and antibiotic tolerant Persisters" however the experimental data shows that the drugs work on populations of cells in models which reduce the growth rate of Mtb. Whilst these experiments show the efficacy of the drugs against POPULATIONS of cells which are tolerant to other anti-TB drugs the experiments do not measure killing of the specialised sub-population of persister cells. To do this the authors would need to perform kill curves to test for changes in the biphasic kill curve (see more about this below). This just need rephrasing so that it doesn't mislead the reader. This also needs to be considered throughout the rest of the document.
We agree that the term "persisters" may not be accurate in this context. We have replaced the term "persisters" by "antibiotic-tolerant, non-replicating mycobacteria" or "non-replicating mycobacteria" throughout the text.
In the abstract the following sentence was amended (Lines 12-14): "The drug combination was bactericidal against replicating and antibiotic-tolerant, nonreplicating mycobacteria, and increased efficacy relative to that of a single drug in a mouse model." In the Introduction section, the following sentences were revised (Lines 50-54): "Hence, with the aim of shortening TB treatment, it is only logical to pursue the development of novel drugs that are active against antibiotic tolerant, non-replicating M. tuberculosis. Although the physiology of non-replicating mycobacteria is not fully understood, bioenergetics is a validated target space (Berube et al, 2019;Gengenbacher et al, 2010;Koul et al, 2008;Rao et al, 2008)." In the Results section, the following sentence was revised (Lines 236-238): "The main limitation of Q203 and related QcrB inhibitors lies in their lack of bactericidal efficacy (Foo et al, 2018;Kalia et al., 2017)." In the Discussion section, the following sentence was revised (Lines 272-276): "We and others have established that oxidative phosphorylation is essential in the maintenance of bioenergetics homeostasis in antibiotic-tolerant hypoxic (Rao et al., 2008) and nutrient-starved (Gengenbacher et al., 2010) M. tuberculosis, opening a scientific rationale to eradicate antibiotic-tolerant, non-replicating subpopulations Hards & Cook, 2018;Kalia et al., 2017;Koul et al., 2008)." (2) It is interesting that the drug combination was more effective against BCG as compared with Mtb. Why do the authors think this is the case? It seems worthy of a line of discussion. And considering this its strange that for the identified resistant strain the authors only investigated BCG. Were the Mtb drug resistant strains also investigated?
The M. bovis BCG Cyt-bcc:aa 3 and Cyt-bd share 100% sequence similarity compared to the M. tuberculosis H37Rv counterparts (Brosch et al., 2007;Lew et al., 2011;Data ref: Garnier T., 2006;Data ref: Lew J. M., 2012). The reason behind the differences in the effectiveness of the drug combination is unclear but could possibly be due to differences in the expression levels of the terminal oxidases in laboratory-adapted M. tuberculosis strains, as described before (Arora et al, 2014).
The M. tuberculosis escape mutants were not investigated because we have lost access to the BSL-3 laboratory in which the mutants were stored. Due to the COVID-19 pandemic, the laboratory is dedicated to SARS-CoV2-related work only for the foreseeable future.
The following sentence was added to the main text (lines 212-214): "Given that the Cyt-bcc:aa 3 and Cyt-bd of M. bovis BCG and M. tuberculosis H37Rv share 100% sequence similarity, escape mutants isolated from M. bovis BCG were selected as representatives in the subsequent studies." (3) For some of the experiments there needs to be some justification on the strain use as data seems to be presented from BCG, Mtb and M. smegmatis in Figure 1.
Referee 2 had a similar concern that was addressed (please see the response to the first major comment of referee #2).
(4) For the experiments where the authors measure the killing efficiency of the drug combinations in the Betts starvation model and a hypoxic model the authors pick one time point. Although there is an inoculum bar there doesn't seem to be a drug free control on figure 4b and 4c whereas there is a control on the growing cells (4a) and there is no Isoniazid control on figure 4a which are important and should be added. Im convinced that the drug combination works on these slow growing drug tolerant cells but without a time course the authors cannot comment on the effect on the persister population.
Figures 4B and C were modified to include the drug free controls.
We usually do not use isoniazid as a positive control in kill-kinetic experiments against replicating mycobacteria because of the rapid emergence of resistance in vitro (Gumbo et al, 2007;Vilchèze et al, 2018). Isoniazid kills mycobacteria rapidly in vitro, usually achieving 3-4 log 10 of killing after 2-4 days of treatment. However, the rapid emergence of resistance to isoniazid result in a resurgence of multiplication from day 5-6 onward (Gumbo et al., 2007;Vilchèze et al., 2018). Given the delayed bactericidal effect of bedaquiline and Q203 (against ΔcydAB strain) (Kalia et al., 2017), we estimated that isoniazid is not an appropriate control drug to be used to compare the bactericidal potency of ND-011992+Q203 against replicating mycobacteria in a 10-day assay. Instead, we used bedaquiline, which is known to be bactericidal against replicating and non-replicating mycobacteria and acting on the same pathway as ND-011992+Q203. Do the data meet the assumptions of the tests (e.g., normal distribution)? Describe any methods used to assess it.
Is there an estimate of variation within each group of data?

Reporting Checklist For Life Sciences Articles (Rev. June 2017)
This checklist is used to ensure good reporting standards and to improve the reproducibility of published results. These guidelines are consistent with the Principles and Guidelines for Reporting Preclinical Research issued by the NIH in 2014. Please follow the journal's authorship guidelines in preparing your manuscript.

B-Statistics and general methods
the assay(s) and method(s) used to carry out the reported observations and measurements an explicit mention of the biological and chemical entity(ies) that are being measured. an explicit mention of the biological and chemical entity(ies) that are altered/varied/perturbed in a controlled manner.
a statement of how many times the experiment shown was independently replicated in the laboratory.
Any descriptions too long for the figure legend should be included in the methods section and/or with the source data.
In the pink boxes below, please ensure that the answers to the following questions are reported in the manuscript itself. Every question should be answered. If the question is not relevant to your research, please write NA (non applicable). We encourage you to include a specific subsection in the methods section for statistics, reagents, animal models and human subjects.
definitions of statistical methods and measures: a description of the sample collection allowing the reader to understand whether the samples represent technical or biological replicates (including how many animals, litters, cultures, etc.).
The data shown in figures should satisfy the following conditions: Source Data should be included to report the data underlying graphs. Please follow the guidelines set out in the author ship guidelines on Data Presentation.
Please fill out these boxes ê (Do not worry if you cannot see all your text once you press return) a specification of the experimental system investigated (eg cell line, species name).
Sample-size calculations were not performed. All experiments were done with replicates and successfully reproduced to demonstrate the consistency of findings.
graphs include clearly labeled error bars for independent experiments and sample sizes. Unless justified, error bars should not be shown for technical replicates. if n< 5, the individual data points from each experiment should be plotted and any statistical test employed should be justified the exact sample size (n) for each experimental group/condition, given as a number, not a range; Each figure caption should contain the following information, for each panel where they are relevant:

Captions
Sample size per group was determined to be n = 4.
All samples were included in the analysis.
Mice were allocated to groups at random.
Samples were considered to be normally distributed.
Yes, the error bars in figures indicate the standard deviation within each group.
For animal studies, mice were allocated to groups at random.
Investigators were not blinded.
For animal studies, no blinding was done.

Data
the data were obtained and processed according to the field's best practice and are presented to reflect the results of the experiments in an accurate and unbiased manner. figure panels include only data points, measurements or observations that can be compared to each other in a scientifically meaningful way.