Transcriptional Inhibition of the F₁F₀-Type ATP Synthase Has Bactericidal Consequences on the Viability of Mycobacteria

INTRODUCTION

There is an urgent need to combat the growing rise and spread of drug-resistant strains of Mycobacterium tuberculosis, the causative agent of tuberculosis and the leading cause of death by a single infectious agent (1). Bedaquiline (BDQ) is a species-specific inhibitor of the mycobacterial F₁F₀-type ATP synthase and is the first antibiotic to receive FDA approval for the treatment of tuberculosis in 40 years (2). Although BDQ is associated with potential cardiac toxicities (1), FDA approval highlights the urgent need for novel therapeutics.

BDQ is characterized by delayed in vitro time-dependent bactericidal activity (3–5). This delayed killing is thought to be the result of the ability of M. tuberculosis to remodel various metabolic pathways to enter and persist in a state of slowed replication that requires only minimal ATP input (3). By binding to the ATP synthase, BDQ perturbs the interaction between the a and c subunits of the ATP synthase, uncoupling ATP synthesis from proton translocation (6). Biochemical studies have shown that BDQ also acts as an H⁺/K⁺ ionophore that, on ATP synthase binding, generates an uncoupled microenvironment, accelerating the rate of respiration and depleting the proton motive force (4, 5, 7, 8). Missense mutations in atpC decrease the rate of ATP synthesis and, in a similar manner to BDQ, accelerate the rate of respiration, resulting in the dysregulation of mycobacterial metabolism (9). Conditional deletion mutants and transposon mutagenesis have established the essentiality of the ATP synthase for mycobacterial growth, highlighting the suitability of the mycobacterial ATP synthase as a drug target (10–13). This current study sought to further explore the essentiality and phenotypic consequences of inhibiting the ATP synthase in mycobacteria. To investigate this, we used CRISPR interference (CRISPRi) to construct transcriptional knockdowns of the ATP synthase operon in M. tuberculosis and Mycobacterium smegmatis.

Mycobacterial CRISPRi utilizes 20 to 25 nucleotide sequences, termed sgRNA, with complementarity to target genes to guide a catalytically inactive Cas9 nuclease from Streptococcus thermophilus CRISPR1 (herein dCas9) to bind target genes (14). Target-bound sgRNA-dCas9 impedes the progression of RNA polymerase, thereby inhibiting transcription (14). sgRNA and dCas9 are expressed from a single integrative plasmid and from anhydrotetracycline (ATc)-inducible promoters (14). The ATP synthase is encoded by an eight-gene operon in M. tuberculosis (Rv1304 to 1311; atpBEFHAGDC). We constructed sgRNA to target the first two genes, atpB and atpE, that are also the targets of BDQ (Fig. 1A and B). There is permissibility in the protospacer-adjacent motif (PAM) sequences targeted by S. thermophilus CRISPR1 dCas9 used in this mycobacterial CRISPRi system, with previous studies identifying 15 variant sequences that achieved >25-fold repression of a luciferase reporter (14). Of the selected sgRNAs, atpB_a, atpB_c, and atpE_a have PAM scores of 9, 2, and 5, respectively. PAM sequences (5′→3′ direction on the template strand) and the targeting region (5′→3′ direction on the nontemplate strand) for selected sgRNAs are shown in Table S1 in the supplemental material. sgRNA-expressing CRISPRi plasmids were constructed and transformed into M. tuberculosis strain mc²6230 (ΔRD1, ΔpanCD) as described in the supplemental material (14, 15). M. tuberculosis strain mc²6230 is an avirulent auxotroph that has been approved for use under BSL2 containment at the University of Otago (15).

• Open in new tab
• Download powerpoint

FIG 1

Transcriptional repression of the ATP synthase in M. tuberculosis using CRISPRi. (A) Organization of the ATP synthase operon in M. tuberculosis. Colored bars above genes indicate approximate locations of sgRNA within the coding sequences of atpB and atpE. Black bars below genes indicate approximate locations of qPCR probes. (B) Schematic of the mycobacterial ATP synthase, highlighting the locations of the a subunit encoded by atpB and the c ring encoded by atpE. (C, D) Transcriptional repression of atpB and atpE in strains of M. tuberculosis expressing atpB- or atpE-targeting sgRNA in the presence of increasing concentrations of ATc. mRNA levels are expressed relative to a strain expressing a nontargeting (NT) sgRNA with 100 ng/ml ATc. Results are means ± SD of technical triplicates. *, P value of < 0.05 as determined by a t test. (E, F) Growth and viability of M. tuberculosis strains expressing atpB- or atpE-targeting sgRNA with various levels of ATc were determined in 96-well plates. Growth (E) was determined after 7 days of incubation, whereas viability (F) was determined after 5 days of incubation. An NT and an mmpL3-targeting sgRNA are included as negative and positive inhibition controls. Results are means ± SD of biological triplicates from a representative experiment (n = 3). (G) M. tuberculosis strains expressing atpB- or atpE-targeting sgRNA were grown with 100 ng/ml of ATc in 10-ml volumes from a starting OD₆₀₀ of 0.005. CFU/ml was determined on indicated days. Results are means ± SD of biological triplicates from a representative experiment (n = 2). (H) M. tuberculosis parental strain mc²6230 was grown in 7H9-OADC-PAN from a starting OD₆₀₀ of 0.005. CFU/ml was determined on indicated days. Results are means ± SD of biological triplicates from a representative experiment (n = 2). CFU/ml data for the NT and atpE_a sgRNAs are taken from panel G. (F, G, H) Dashed horizontal lines, upper and lower levels of detection; Inoc, CFU/ml at inoculation (time 0).

The strength of transcriptional repression by each sgRNA was determined by growing cultures in Middlebrook 7H9 broth supplemented with OADC (0.005% oleic acid, 0.5% bovine serum albumin, 0.2% dextrose, 0.085% catalase), 0.05% tyloxapol (Sigma), 25 μg/ml pantothenic acid (PAN), and 25 μg/ml kanamycin (KAN) from a starting optical density at 600 nm (OD₆₀₀) of 0.1. Cultures were induced with ATc and grown for 3 days, after which RNA was extracted, cDNA was synthesized, and quantitative PCR (qPCR) was performed as described in the supplemental material. With 100 ng/ml of ATc, the sgRNAs atpB_a and atpB_c repressed the expression of atpB by 78 ± 8- and 78 ± 16-fold, respectively (Fig. 1C). With 10 ng/ml of ATc, atpB_a and atpB_c had 5-fold repression of atpB expression. With 100 ng/ml of ATc, atpE_a that targets the downstream atpE had 4-fold repression of atpB expression (Fig. 1C). With 100 ng/ml of ATc, atpE_a reduced expression of atpE by 46 ± 3-fold (Fig. 1D). With 100 ng/ml of ATc, atpB_a and atpB_c that target atpB also reduced the expression of atpE, causing 23 ± 4- and 19 ± 3-fold repression, respectively (Fig. 1D). This is consistent with mycobacterial CRISPRi having polar effects on downstream cotranscribed genes (14). In conclusion, CRISPRi is able to transcriptionally repress the expression of the ATP synthase operon in M. tuberculosis.

The phenotypic consequences of inhibiting the transcription of the mycobacterial ATP synthase operon were determined in a 96-well-plate format following previously described protocols (15). Briefly, cultures were inoculated at a starting OD₆₀₀ of 0.005 in 7H9 broth supplemented with OADC-KAN-PAN along a 3-fold dilution gradient of ATc in a total volume of 100 μl. Plates were incubated at 37°C without agitation. Essentiality screens were performed after 7 days incubation by measuring OD₆₀₀ in a Varioskan LUX microplate reader and determining the MIC of ATc using a nonlinear fitting of data to the Gompertz equation (15, 16). Effects on bacterial viability were obtained by determining the change in CFU/ml after 5 days incubation because previous work had shown that maximum killing before the emergence of ATc-nonresponsive mutants was typically achieved by this time (15). All ATP synthase-targeting sgRNAs prevented bacterial growth, with ATc MICs of between 4 and 12 ng/ml, confirming the essentiality of the mycobacterial ATP synthase (Fig. 1E). Unexpectedly, we observed various phenotypes for each sgRNA on bacterial viability. The sgRNA atpB_a slowed growth at 100 and 300 ng/ml ATc, causing a <1-log₁₀ increase after 5 days (Fig. 1F). The sgRNA atpB_c caused a <1-log₁₀-CFU/ml reduction at ATc concentrations between 10 and 100 ng/ml but a 1.7 ± 0.4-log₁₀ reduction at 300 ng/ml ATc (Fig. 1F). The sgRNA atpE_a was bactericidal after 5 days, causing a 1.0 ± 0.04-log₁₀-CFU/ml reduction with 10 ng/ml ATc and a >2-log₁₀-CFU/ml reduction at ATc concentrations of >33 ng/ml (Fig. 1F). The level of bacterial killing achieved by atpE_a was comparable to that achieved by an sgRNA targeting mmpL3, a known bactericidal target (15) (Fig. 1F). To validate these results, time-kill assays were performed by inoculating cultures at a starting OD₆₀₀ of 0.005 into 10 ml 7H9 broth supplemented with OADC-KAN-PAN with 100 ng/ml ATc as described in the supplemental material. The sgRNA atpB_a caused a slowing of bacterial growth over a 9-day period, allowing for a 1-log-CFU/ml increase by day 9 (Fig. 1G). atpB_c caused a 1.1 ± 0.3-log₁₀-CFU/ml reduction by day 9, whereas atpE_a was bactericidal, causing a 2.8 ± 0.3-log₁₀-CFU/ml reduction by day 5 (Fig. 1G). Killing by atpE_a was comparable to the 3.3 ± 0.4-log₁₀-CFU/ml reduction caused by the inhibition of mmpL3 (Fig. 1G). Both atpE_a and mmpL3 selected for escape mutants, as observed by the increase in CFU/ml after maximum killing was achieved (Fig. 1G). The variation in phenotypes associated with each sgRNA was unlikely to be a result of off-targeting effects, due to the more restrictive seven-nucleotide PAM sequence of S. thermophilus CRISPR1 dCas9 (17). Previous work examining pooled CRISPRi arrays in M. smegmatis also failed to identify “bad seed” effects (17). Variation in PAM score of each guide may influence the rate of dCas9 unbinding from the target gene over a prolonged period of incubation as used in the phenotypic assay compared to the transcriptional analysis, with the growth-slowing atpB_a having a weaker PAM score than the bacteriostatic atpB_c (PAM score, 9 versus 2, respectively). It is also possible that the greater repression of atpE achieved by atpE_a than by atpB_a or atpB_c is responsible for the bactericidal phenotype. To compare killing between transcriptional and chemical inhibition of the ATP synthase, time-kill assays of BDQ against the parental M. tuberculosis mc²6230 strain were performed as described above. With 30× MIC of BDQ, there was a lack of killing over the first 5 days, followed by a 0.5-log-CFU/ml reduction by day 7 (Fig. 1H). These results were consistent with experiments in M. tuberculosis strain H37Rv (3). Furthermore, there was no difference in killing between 30× and 300× MIC of BDQ over this time course (3). Direct comparison of time-kill assays for transcriptional and chemical inhibition of the ATP synthase further highlighted the increased rate of killing achieved by transcriptional inhibition (Fig. 1H). In conclusion, the transcriptional inhibition of the ATP synthase can have bactericidal consequences on the viability of M. tuberculosis.

To validate the bactericidal consequences of transcriptionally inhibiting the mycobacterial ATP synthase, we constructed CRISPRi plasmids that targeted atpB and atpE in M. smegmatis, a faster-growing mycobacterial species. As with M. tuberculosis, BDQ has a time-dependent killing of M. smegmatis (4), and genes encoding the ATP synthase are arranged in an operon (MSMEG4941-4932), with atpB and atpE being the first and second genes. PAM sequences, PAM score, and targeting regions of selected sgRNAs are shown in Table S1. With 100 ng/ml of ATc, sgRNAs atpB_a and atpB_b repressed atpB expression by 150 ± 47- and 1,095 ± 311-fold, respectively (Fig. 2A). sgRNA atpE_a repressed atpE expression by 1,115 ± 433-fold (Fig. 2B). The effects of these sgRNAs on bacterial growth and viability were determined in a 96-well-plate format, with culture inoculated at an OD₆₀₀ of 0.005 into Hartmans-de Bont minimal medium with succinate as a sole carbon source (18). ATc was prepared as a dilution gradient along the y axis. OD₆₀₀ and CFU/ml were determined after 26 h incubation at 37°C with agitation (200 rpm). Consistent with M. tuberculosis, all sgRNAs targeting the ATP synthase inhibited bacterial growth, with an MIC against ATc of ∼0.01 to 0.07 ng/ml (Fig. 2C). The reduced ATc MIC was due to the increased efficacy of ATc in M. smegmatis in inactivating the tetracycline repressor pools required to activate transcription (14). Consistent with the bactericidal phenotype of atpE_a in M. tuberculosis, all sgRNA targeting atpE and atpB in M. smegmatis had bactericidal consequences on cellular viability, with a >2-log₁₀-CFU/ml reduction at ATc concentrations between 1 and 100 ng/ml (Fig. 2D). The level of killing was comparable to that achieved by an sgRNA targeting mmpL3 (Fig. 2D). We hypothesize that the consistency of bactericidal phenotypes among atpB- and atpE-targeting sgRNAs in M. smegmatis is a result of the stronger transcriptional repression. Combined, results from M. tuberculosis and M. smegmatis demonstrate that the transcriptional repression of the ATP synthase has potential for bactericidal consequences on the viability of mycobacteria.

• Open in new tab
• Download powerpoint

FIG 2

Transcriptional repression of the ATP synthase in M. smegmatis using CRISPRi. (A) Expression levels of atpB in strains of M. smegmatis expressing either atpB_a or atpB_b sgRNA. (B) Expression levels of atpE in a strain of M. smegmatis expressing the sgRNA atpE_a. In both panels: boxes above genes indicate approximate locations of sgRNA within the coding sequences of atpB and atpE; black bars below genes indicate approximate locations of qPCR probes. Results are means ± SD of technical triplicates. (C, D) Growth and viability of M. smegmatis strains expressing atpB- or atpE-targeting sgRNA with various levels of ATc were determined in 96-well plates. Growth (C) and viability (D) were determined after 26 h. An NT and an mmpL3-targeting sgRNA are included as negative and positive inhibition controls. Results are means ± SD of biological duplicates from a representative experiment (n = 2). Dashed horizontal lines, lower level of detection; Inoc, CFU/ml at inoculation (time 0).

Our results highlight a clear contrast between the consequences of transcriptional and chemical inhibition (i.e., BDQ) of the mycobacterial ATP synthase on mycobacterial viability. Conditions in which the ATP synthase is partially repressed should increase its sensitivity to inhibition by BDQ. Initially, in a 96-well-plate format, we tested the sensitivity to BDQ of the phenotypically weakest sgRNA (i.e., atpB_a) with increasing concentrations of ATc. The 96-well plates were prepared as described in the supplemental material, and MIC and CFU/ml were determined after 10 days incubation at 37°C without agitation. The addition of increasing ATc concentrations reduced the BDQ MIC from 0.9 μM when uninduced to 0.03 and 0.02 μM with 10 or 100 ng/ml of ATc, respectively (Fig. 3A). Addition of 10 ng/ml of ATc also resulted in a bacteriostatic phenotype, with a 0.5- and 0.3-log₁₀-CFU/ml reduction compared with the inoculating CFU/ml at one-third and one-ninth the MIC of BDQ (Fig. 3B). The addition of 100 ng/ml of ATc resulted in 1.1- and 0.9-log₁₀ CFU/ml reductions compared with the inoculating CFU/ml at one-third and one-ninth the MIC of BDQ, respectively (Fig. 3B). There was no change in MIC for a nontargeting sgRNA (Fig. 3C). Time-kill assays were performed to further explore the increased sensitivity of the atpB transcriptional knockdown to BDQ. Induction of atpB_a alone delayed bacterial growth (Fig. 3D). Strains of M. tuberculosis expressing atpB_a sgRNA were killed by concentrations of BDQ below the MIC, with a 3-log₁₀-CFU/ml reduction by day 21, while uninduced strains grew at an impaired rate (Fig. 3D). In conclusion, the transcriptional repression of the mycobacterial ATP synthase reduces the concentration of BDQ required to kill M. tuberculosis.

• Open in new tab
• Download powerpoint

FIG 3

Transcriptional inhibition of atpB sensitizes M. tuberculosis to BDQ. M. tuberculosis expressing atpB-targeting (i.e., atpB_a) (A and B) or NT (C) sgRNA was grown in 96-well plates in the presence of 0, 10, or 100 ng/ml ATc and a 9-point 3-fold dilution gradient of BDQ. OD₆₀₀ (A and C) and CFU/ml (B) were determined on day 10. (B) Dashed horizontal lines represent upper and lower levels of detection; DMSO, dimethyl sulfoxide-treated cells (i.e., no BDQ) correspond to the day 10 CFU/ml reached by M. tuberculosis expressing the atpB_a sgRNA induced with 0, 10, or 100 ng/ml ATc. Results are means ± SD of biological duplicates from a representative experiment (n = 3). (D) M. tuberculosis strains expressing the atpB-targeting sgRNA atpB_a were grown with or without 100 ng/ml of ATc in 10-ml volumes from a starting OD₆₀₀ of 0.005. Cultures were grown with or without 0.1 μM BDQ (MIC, 0.3 μM). CFU/ml was determined on stated days. Dashed horizontal lines represent upper and lower levels of detection. Results are means ± SD of biological triplicates from a representative experiment (n = 2). Inoc, CFU/ml at inoculation (time 0).

Here, we describe that the transcriptional inhibition of the ATP synthase operon has bactericidal consequences on the viability of both M. tuberculosis and M. smegmatis. This is in stark contrast to BDQ, a chemical inhibitor of mycobacterial ATP synthase that has a slow time-dependent killing. It is possible that these phenotypic differences are a result of the differences between transcriptional and chemical inhibition. For example, transcriptional inhibition will have effects on both enzyme catalysis and complex formation, whereas BDQ only affects enzyme catalysis in addition to its activity as an uncoupler (4, 7, 19, 20). Despite this, the results support continued drug discovery efforts to identify and develop novel mycobacterial ATP synthase inhibitors that have a rapid bactericidal activity.