RNase HI Depletion Strongly Potentiates Cell Killing by Rifampicin in Mycobacteria

To quantify R-loops in the RNase HI depletion strains, we isolated total nucleic acid from cultures of both ΔrnhA and ΔrnhC strains grown to mid-log phase and quantified the relative amounts of R-loops in each, using Southwestern blots with a monoclonal antibody specific for RNA/DNA hybrids (24). The specificity of the antibody for RNA/DNA hybrids was confirmed by subjecting samples to RNase HI digestion prior to detection and noting the loss of signal in control spots (Fig. 2A). R-loops were detected at a low level in the parental strain and accumulated in the ΔrnhA mutant by a factor of ~2-fold and in the ΔrnhC mutant by a factor of ~11-fold (Fig. 2B). This indicated that both RnhA and RnhC act as part of a protective system to remove R-loops in replicating and actively transcribing cells. In both cases, the accumulation of R-loops also indicated that neither the remaining RNase HI isoform, nor other potential hybrid removal mechanisms such as the DNA helicase RecG (25), are able to compensate fully for the loss of either RNase HI enzyme.

To confirm that loss of RNase HI activity was responsible for the observed increased abundance of R-loops in the ΔrnhA and ΔrnhC strains, we tested whether either rnhA or M. tuberculosis rnhC provided in trans could complement the ΔrnhA and the ΔrnhC mutants, respectively (Fig. 2). As expected, complementation with the plasmid-borne rnhA gene reduced the R-loop signal in ΔrnhA to a wild-type level (Fig. 2A). In contrast, introduction of a plasmid carrying rnhC into the ΔrnhC strain conferred a 3-fold increase of R-loops over ΔrnhC without plasmid. To determine whether the R-loop signal in the complemented strain might be plasmid-borne, we separated total nucleic acid on an agarose gel, transferred it to a nylon membrane, and carried out Southwestern blotting as before (Fig. 2C to E). Only the ΔrnhC strain showed the presence of R-loops, running as a smear of smaller fragments below the main band of chromosomal DNA. No R-loops were detected in the complemented strain, even after very long exposures (Fig. 2E), consistent with a reporter-based indication of restoration of normal topology by provision of rnhC in trans (see below). A dot blot with the same nucleic acid preparation used for the agarose gel electrophoresis (Fig. 2D) recapitulated the previous findings of excess R-loops in the ΔrnhC complemented strain, indicating that subjecting the sample to gel electrophoresis abolished the R-loop signal in the complemented strain but not in the ΔrnhC strain, showing that the R-loop signal observed in the dot blot of the complemented strain consisted of less stable or transient R-loops. R-loops that form in suboptimal regions of DNA under conditions of high topological stress, such as in small circular molecules undergoing active transcription, dissociate readily when topological stress is removed (26). Notably, the RNA/DNA hybrid signal corresponded to low molecular weight nucleic acid migrating below the bulk of the chromosomal DNA, indicating that the hybrid molecules were more sensitive to breakage than nonhybridized DNA, in keeping with their proposed role in generating double-stranded breaks. No low-molecular-weight nucleic acid was visible in the complemented strain, supporting the nonpathological nature of any R-loops captured in the cognate dot blot of the complemented strain. This conclusion is also congruent with subsequent experiments which showed full complementation with the complementation plasmid.

Persistent R-loops, but not transient R-loops, could be expected to affect topology homeostasis. In light of the previous results, and since RnhC appeared to be the dominant contributor to R-loop resolution in M. smegmatis, we decided to probe topology homeostasis in the ΔrnhC mutant. In transcriptionally active cells, negative supercoils accumulate in the wake of the RNA polymerase, and positive supercoils accumulate ahead of it, a phenomenon known as the “twin domain hypothesis” (27, 28) (Fig. 1), which requires the activities of both topoisomerase I and DNA gyrase I to resolve (29, 30). Positive supercoils accumulate, in particular, between convergently oriented transcription units and also between converging replication complexes and transcription complexes, with negative supercoils accumulating upstream. Since negative supercoiling in the wake of the polymerases can promote R-loop formation, trapping the RNA polymerase on the DNA and providing a barrier to supercoil diffusion, we expected that there might be an increased demand in the ΔrnhC mutant for either topoisomerase I (which resolves negative supercoiling) or DNA gyrase I (which resolves positive supercoiling), or both, due to reduced ability to remove these R-loop barriers. Both the DNA gyrase and topoisomerase I promoters in M. smegmatis are coupled directly to topological stresses experienced by the chromosome (responding by increasing expression up to 4-fold and 2-fold, respectively), as RNA polymerase binding elements (−35 and −10 sites) in each promoter adopt optimal conformations for transcription only if the DNA is either overwound (DNA gyrase) (31) or underwound (topoisomerase I) (32). An increase in expression of both promoters would therefore indicate an increased demand consistent with the presence of persistent R-loops (i.e., twin domains of positive and negative supercoiling generated by transcription but unable to be resolved by diffusion). We fused these promoters to a lacZ reporter gene and integrated these reporter cassettes into the attB site of the chromosomes of both wild-type and ΔrnhC M. smegmatis strains. Reporter activity from gyrase and topoisomerase promoters was increased significantly in the ΔrnhC strain compared to the wild type, by 3-fold (P = 0.002) and 1.7-fold (P = 0.02), respectively (Fig. 3). Notably, these increases in activity were abolished in the complemented ΔrnhC strain, suggesting first, that the ΔrnhC strain was experiencing increased levels of both positive and negative supercoiling and, second, that this topological stress was effectively relieved by the complementing rnhC gene provided in trans.

Aberrant supercoiling status can affect control of plasmid replication mechanisms, which in turn can alter plasmid copy number (10, 33–35). Notably, the pAL5000 origin of replication has been shown to be sensitive to phasing alterations which can either increase or decrease copy number (36). Read number comparisons from whole-genome sequencing (WGS) have been used to establish relative gene dosage (37), so we extended this concept to infer plasmid copy numbers relative to the chromosome. We carried out WGS of strains grown to mid-log phase to determine copy numbers of the pAL5000-based complementing plasmids in the ΔrnhA or ΔrnhC deleted strains (Table 1). The HygR-encoding vector backbone (pOLYG) was present at a copy number of 3.7 per genome in the wild type, which is consistent with reported values of 3 to 5 per cell (38), indicating that this method was indeed suitable for determining plasmid copy number. In contrast to the low copy number of the HygR-encoding plasmid in wild-type cells, the copy number of the KmR-encoding plasmid (pBBK) was present in wild-type cells at around 13 copies per genome, 4-fold greater than in the HygR-encoding plasmid.

Inspection of the plasmid sequence revealed a 12-bp deletion in the N terminus of the repB gene in the KmR-encoding vector backbone (see Fig. S1 in the supplemental material), leading to the deletion of one set of three tetrapeptide repeats. RepB belongs to the sigma factor family of proteins and is essential for plasmid replication (39). The C-terminal part of RepB binds to sites flanking the origin of replication, while the N terminus of RepB interacts with the C terminus of the RepA primase/polymerase and aligns it to bind to a highly conserved AT-rich sequence important for replication initiation (39). In previous work, it was determined that as little as a 3-bp deletion in the C terminus of RepA was sufficient to raise the copy number of the pHIGH100 replicon by 8-fold (40), suggesting that copy number can be affected by either alterations in RepA affinity for RepB or by a change in the relative alignment of RepA to the initiation region. A phase change equivalent to one helical turn of DNA in this region (+10 bp) increased the copy number 2-fold (36). The 12- bp deletion observed in RepB might therefore be responsible for the higher basal copy number in wild-type cells via a similar route, but investigation of the precise mechanism was not pursued further in this study.

Plasmid copy number was increased ~6-fold over wild-type levels to 21 per genome in the ΔrnhA mutant. In the ΔrnhC strain, the relative copy number of this plasmid was raised 2-fold over the wild-type level, to 28 per genome. Both the ΔrnhA and ΔrnhC strains therefore showed an increase in plasmid copy number over wild-type levels, although the initial effect on copy number was greater in ΔrnhA than in ΔrnhC. This suggested an involvement of either topology or R-loop homeostasis in copy number control. However, complementation with plasmid-borne rnhA or rnhC did not reduce the plasmid copy number to wild-type levels; instead, the trend was to increase it slightly. WGS confirmed that the rnh-deleted strains were isogenic with their respective complemented strains, eliminating compensatory mutations in host genes as a potential contributor to this effect. However, plasmid supercoiling status can differ markedly from chromosomal supercoiling status since host factors can be titrated to the chromosome (34). The observation of high levels of transient R-loops in the ΔrnhC-complemented strain might indicate higher levels of negative supercoiling or R-loops in the complementing plasmids, resulting in the sustained elevation of the copy number; however an investigation of the molecular basis for these high copy numbers was not undertaken.

Regardless of the precise mechanism, since R-loop accumulation could be complemented in trans by these plasmids in both the ΔrnhA and ΔrnhC strains, and the chromosomal DNA topology homeostasis was fully restored in ΔrnhC, we concluded that the plasmid copy number was well suited to the cellular requirements for RNase HI activity and that the increased copy number of the plasmid in the complemented strains might even reflect a useful cellular feedback mechanism to titrate the level of RNase HI appropriately. For these reasons, subsequent experiments were carried out using these constructs and strains.

Since transcription is a prerequisite for R-loop formation, we hypothesized that transcriptional inhibition would reduce the number of R-loops in the cell. Rifampicin, which binds to the RpoB subunit of RNA polymerase, inhibits transcription at the initiation step. Importantly, it does not affect RNA polymerase complexes that have added more than three ribonucleotides from proceeding to active transcription (41) and thus is not expected to uncouple transcription and translation in actively transcribing polymerases. Rifampicin has been used to reverse high levels of negative supercoiling in plasmids arising from transcription (42) and to reverse effects arising from R-loop accumulation (43, 44).

We exposed mid-log phase cultures of wild-type and ΔrnhC strains of M. smegmatis to rifampicin for 1 h before harvesting the cells and quantitating the R-loops present in them. Rifampicin at a concentration equivalent to the MIC₉₀ value for M. smegmatis (4.2 μM or 3.5 μg mL⁻¹) abolished R-loops, a result consistent with complete inhibition of transcription (Fig. 4). Surprisingly, both the wild-type and ΔrnhC strains contained significantly increased amounts of R-loops when exposed to a lower concentration of rifampicin (equivalent to the MIC₅₀ value; 1.2 μM or 1 μg mL⁻¹). A pronounced response to this low rifampicin exposure was shown by the ΔrnhC strain, with the number of R-loops increasing by approximately 3-fold compared to the nontreated ΔrnhC strain. Strikingly, even at a rifampicin concentration equivalent to the MIC₂₅ (0.6 μM or 0.5 μg mL⁻¹), more R-loops accumulated in ΔrnhC cells than in untreated cells, indicating an overload of the cellular systems required for R-loop resolution. This also suggested that the low number of R-loops observed in wild-type cells at this rifampicin concentration was not due to a lack of R-loop formation but, rather, was reflective of efficient resolution by RNase HI. Notably, R-loop resolution is clearly impaired even in wild-type cells exposed to rifampicin at the MIC₅₀. These results indicate that under these conditions, sublethal concentrations of rifampicin cause R-loop accumulation, a phenomenon that has not previously been described and is opposite to the effect of a lethal concentration of rifampicin.

To better understand the cellular consequences of rifampicin-induced R-loop accumulation, we carried out a dose-response assay to assess the extent to which RNase HI contributes to the survival of cells exposed to rifampicin. We found that wild-type M. smegmatis is inhibited as expected by previously established concentrations of rifampicin (45, 46). In contrast, loss of rnhC strongly sensitized the rnhC knockout strain to rifampicin, reducing the MIC₅₀ almost 100-fold, from 1.2 μM to 18 nM (Fig. 5A). This sensitivity was fully complemented by the provision of rnhC in trans. To confirm that this remarkable loss of viability under rifampicin stress is due to the loss of the RNase HI domain rather than the CobC domain of RnhC, which is involved in vitamin B₁₂ biosynthesis (20), the assay was repeated in the presence of vitamin B₁₂ in the growth medium. Supplementation with vitamin B₁₂ altered the MIC₅₀ slightly, but not significantly (P = 0.3), from 18 nM to 30 nM (Fig. 5A), indicating that the rifampicin sensitization was due to the depletion of the RNase HI function of RnhC.

As the ΔrnhA strain of M. smegmatis showed a smaller accumulation of R-loops than the ΔrnhC strain (Fig. 2), we asked whether the loss of RnhA could also confer increased sensitivity to rifampicin. M. smegmatis ΔrnhA showed an approximately 3-fold increase in sensitivity to rifampicin compared to the wild type, which was fully complemented by the provision of rnhA in trans (Fig. 5B). Together, these findings indicated that depletion of RNase HI activity was responsible for the increased sensitivity of M. smegmatis to transcriptional inhibition and that loss of RnhA had a smaller contribution to this phenotype than the loss of RnhC. In fact, it was remarkable that such an apparently small contribution to R-loop metabolism still conferred synergy with rifampicin, underscoring the vulnerability of mycobacterial cells to even small amounts of RNase HI depletion.

R-loop formation can be affected by uncoupling transcription and translation (47, 48) or by altering DNA supercoiling in the genome (5, 49). Rifampicin treatment does not uncouple transcription and translation, as it does not bind to polymerases that have added more than 3 ribonucleotides. Therefore, we tested the sensitivity of the M. smegmatis ΔrnhC strain to streptomycin (an inhibitor of translation), amsacrine (a topoisomerase I inhibitor), and moxifloxacin (a DNA gyrase inhibitor) (Fig. 1). As a control, we also investigated the sensitivity of the ΔrnhC strain to isoniazid, which disrupts mycobacterial cell wall synthesis by inhibiting mycolic acid synthesis (50).

The loss of the rnhC gene significantly reduced the MIC₅₀ for streptomycin, from 31 nM in the wild-type to 17 nM in the ΔrnhC strain (P = 0.02), and that for moxifloxacin, from 194 nM to 62 nM (Fig. 6A and B) (P = 0.001). In contrast, the loss of rnhC did not result in a significant increase in the sensitivity of M. smegmatis to amsacrine (MIC₅₀, 160 μM for the wild type and 140 μM for the ΔrnhC strain; P = 0.8) (Fig. 6C). As expected, the loss of the rnhC gene did not significantly affect the MIC₅₀ to isoniazid (101 μM for the wild type and 80 μM for the knockout strain, P = 0.2) (Fig. 6D).

Taken together, these results suggest that RNase HI inhibition would improve the efficacy of rifampicin, streptomycin, and moxifloxacin in a synergistic manner. In all cases, the increase in efficacy arising from RNase HI depletion was more pronounced at sub-MICs of antibiotic, which is consistent with a requirement for some transcription to occur for the effect to be seen.

The enhanced antibiotic activity observed when RNase HI activity is depleted in M. smegmatis indicated that RNase HI inhibition is likely to be compatible with existing antibiotics used for M. tuberculosis therapy. Effective RNase HI inhibitors would be of particular interest as synergistic partners for rifampicin, which is a key first-line antibiotic used to treat tuberculosis. Rifampicin is a powerful inducer of hepatic cytochrome P450 enzymes, which leads to antagonistic interactions with many other drugs (51). Hence, any drug that sensitizes M. tuberculosis to rifampicin also has the potential to improve therapeutic outcomes by decreasing the dose of rifampicin required.

Recently, specific inhibition of HIV RNase HI was reported (52–54), preventing viral replication in cells and supporting the possibility that small-molecule compounds could be similarly discovered for bacterial RNase HIs. The drug discovery effort against the RNase HI activity of HIV-1 reverse transcriptase has identified promiscuous chemical scaffolds that not only have activity on HIV RNase HI in vitro, but also have activity on E. coli RNase HI or human RNase HI. We therefore selected a small library of 33 inhibitors of the RNase H activity of HIV-1 reverse transcriptase (Table S4) which we screened for activity against recombinant M. tuberculosis RNase HI. Ten compounds were found to inhibit M. tuberculosis RNase HI at a concentration of less than 100 μM (Table S4).

The 10 effective in vitro inhibitors of M. tuberculosis RNase HI were then tested for their antimicrobial activity against M. tuberculosis in a growth inhibition assay. All showed weak or no growth inhibition of M. tuberculosis when tested alone, which is consistent with their modest in vitro inhibition of RNase HI. Alternative explanations for this could be a lack of penetrance or off-target activity in vivo. However, we hypothesized that on-target inhibition of M. tuberculosis RNase HI in vivo would phenocopy the genetic depletion of RNase HI observed in M. smegmatis by increasing the sensitivity of M. tuberculosis to sub-MIC₅₀ concentrations of rifampicin. Four compounds (NSC353720, NSC600285, NSC18806, and NSC99726) significantly potentiated killing of M. tuberculosis by rifampicin at a sub-MIC, indicating that they are likely to inhibit RNase HI in vivo (Fig. 7A and Table 2). In vitro dose-response curves against M. tuberculosis RNase HI show that NSC353720 (IC₅₀, 14 μM) was the most potent inhibitor, while NSC99726 was the weakest inhibitor (IC₅₀, 45 μM) of this compound set (Fig. 7B and Table 2).

The strains and plasmids used in this study are listed in Tables S1 and S2. E. coli strains Top10 and BL21(DE3)pLysS, used for cloning and protein expression, respectively, were grown in lysogeny broth (LB) and/or on agar at 37°C or 28°C in the presence of the appropriate antibiotic(s) at the following concentrations: 50 μg mL⁻¹ kanamycin (GoldBio), 100 μg mL⁻¹ ampicillin (GoldBio), 200 μg mL⁻¹ hygromycin (Thermo Fisher), 5 μg mL⁻¹ gentamicin (GoldBio) and 35 μg mL⁻¹ chloramphenicol (Sigma-Aldrich). M. smegmatis mc²155 and M. tuberculosis H37Rv were grown at 37°C in Middlebrook 7H9 broth (BD) or on Middlebrook 7H10 agar (BD) supplemented with 10% albumin-dextrose-saline, 0.2% glycerol, and 0.05% Tween 80 (Sigma) in the presence of appropriate antibiotics at the following concentrations: 25 μg mL⁻¹ kanamycin, 50 μg mL⁻¹ hygromycin, and 5 μg mL⁻¹ gentamicin.

Plasmids were constructed using standard cloning techniques. Inactivation of rnh alleles was carried out by homologous recombination with suicide plasmids (85) containing alleles inactivated by insertion of either a hygromycin antibiotic resistance cassette (ΔrnhC) or kanamycin resistance cassette (ΔrnhA). Allele replacement was confirmed initially by Southern blotting and subsequently by whole-genome sequence analysis (data deposited as BioProject no. PRJNA826235). Bioluminescent strains were constructed by transformation with the integrating plasmid pMV306G13+Lux, which carries the luxABCDE operon (86). M. smegmatis rnhA and M. tuberculosis rnhC genes for complementation studies were cloned into the pAL5000-based replicating plasmids (87) pOLYG (HygR) and pBBK (KmR), respectively, to make prnhA (HygR) and prnhC (KmR) (88). The M. smegmatis rnhA gene was cloned with 350 bp of the upstream region containing its own promoter (89), and in pBBK, M. tuberculosis rnhC was expressed from the well-characterized leaderless Ag85 promoter (90).

Drug dose-response curves were plotted, and 50% inhibitory concentration (IC₅₀) and MIC₅₀ determinations were carried out using nonlinear regression analysis. Student’s unpaired t test was used to test for significance throughout. All plots and calculations were carried out using GraphPad Prism 7.

Bacterial strains were grown as described above and harvested at mid-log phase (optical density at 600 nm (OD₆₀₀), 1.5). Nucleic acids were extracted from 20-mL culture aliquots using a modified phenol-chloroform technique (91): following lysozyme treatment, RNase A (10 μg/mL) was added to degrade single-stranded RNA, and small RNAs were precipitated by 30% (wt/vol) PEG 8000/30 mM MgCl₂. Nucleic acid was quantified using a spectrofluorometric method using QuantiFluor double-stranded DNA (dsDNA; Thermo Fisher). Nucleic acid (400 ng) was either separated by electrophoresis on a 1% agarose gel prior to transfer or was spotted directly onto Amersham Hybond N+ nylon membrane (GE Healthcare Lifesciences). The nucleic acid was air-dried and UV cross-linked to the membrane for 3 min using a transilluminator. Control aliquots were preincubated with 10 U of E. coli RNase HI (New England Biolabs [NEB]) for 2 h at 37°C before spotting. The membrane was blocked in TBST (20 mM Tris, pH 7.4, 150 mM NaCl, and 0.1% [vol/vol] Tween 20) containing 5% (wt/vol) milk powder at room temperature for 1 h, which was then removed and replaced with fresh TBST containing 5% (wt/vol) milk powder and S9.6 anti-DNA/RNA hybrid antibody (Kerafast) at 1 μg/mL and incubated at 4°C overnight. The following day, the membrane was washed (3 times for 5 min) with fresh TBST and finally incubated with a 1:1,000 dilution of horseradish peroxidase (HRP)-conjugated anti-mouse secondary antibody (Invitrogen) in TBST for 1 h at room temperature. Unbound secondary antibody was removed by washing (3 times for 5 min) with fresh TBST, and the blot was developed with West Pico PLUS chemiluminescent substrate (Thermo Fisher) according to the manufacturer’s instructions. Chemiluminescence was detected using an Amersham Imager 600, and dot intensity was quantified using the Image Lab software suite (Bio-Rad).

Cultures of M. smegmatis mc²155 and ΔrnhC were grown as described above and harvested at mid-log phase. Optical density was used for normalization of β-galactosidase activity. The assay was carried out according to Miller with variations (92, 93). Briefly, aliquots (2 mL) from each culture were harvested and resuspended in 600 μL of Z-buffer (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, 50 mM β-mercaptoethanol). Cells were permeabilized by adding 100 μL of chloroform and 50 μL of 0.1% SDS, and the mixture was incubated at 28°C for 5 min. Then 200 μL of ONPG (o-nitrophenyl-β-d-galactopyranoside; 4 mg/mL) was added to the mixture, which was incubated at 28°C until sufficient yellow color developed, and the time of incubation was recorded. The reaction was stopped by the addition of 500 μL of 1 M Na₂CO₃ and then the mixture was centrifuged at 10,000 × g for 2 min. The OD₄₂₀ of the supernatant was measured, and the β-galactosidase activity was expressed in Miller units using the following equation: OD₄₂₀ × 1,000/t × v × OD₆₀₀, where t is the incubation time (min) and v is the volume of culture (mL).

(i) M. smegmatis. Two different methods were used to assess growth inhibition. (i) Agar proportion method: M. smegmatis strains were grown and harvested at mid-log phase (OD₆₀₀, 1 to 1.5). A single-cell suspension was prepared by centrifugation of the culture at 2,000 × g for 10 min to pellet clumps, the supernatant was collected, and 20% glycerol was added. Aliquots were snap-frozen in liquid nitrogen and stored at −80°C. For killing assays, 150 CFU were plated in Middlebrook 7H10 agar plates containing 2-fold serial dilutions of rifampicin and isoniazid and incubated at 37°C for 3 days. (ii) Bioluminescence reporter assay: bacterial strains carrying pMV306G13+Lux were grown as described above to mid-log phase (OD₆₀₀, 1.5) and diluted 30-fold to an OD₆₀₀ of 0.05. All assays were set up in 96-well plates in a total volume of 100 μL, where 50 μL of the diluted culture was added to 50 μL of 2-fold serially diluted antibiotics in Middlebrook 7H9 medium. Two controls were added: bacteria only and medium only. The plates were sealed and incubated at 37°C for 18 h. Bioluminescence was quantified using a SpectraMax iD3 plate reader. Bacterial survival at each concentration was calculated based on the ratio of the relative light units (RLU) in the presence or absence of the inhibitor.

(ii) M. tuberculosis. M. tuberculosis mc²6230 (ΔRD1 ΔpanCD) (94) was grown 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, and 25 μg mL⁻¹ pantothenic acid (PAN) at 37°C. The assay was carried out in 96-well plates in a total volume of 100 μL. The cultures were diluted to an OD₆₀₀ of 0.05, and 50 μL of the diluted culture was added to 50 μL of 2-fold serially diluted inhibitor in Middlebrook 7H9 medium. The plates were incubated without shaking at 37°C for 7 days before the MICs that inhibited 90% of growth (MIC₉₀) were determined from the visual presence or absence of growth in biological triplicates. The fractional inhibitory concentration index (FICI) was determined for RNase HI inhibitors in combination with rifampicin by the addition of a sub-MIC of rifampicin (5 nM) to the MIC assay. FICI values were calculated using the equation FICI = FIC_A + FIC_B = (C_A/MIC_A) + (C_B/MIC_B), where MIC_A and MIC_B are the MICs of drugs A and B alone, and C_A and C_B are the concentrations of the drugs in combination.

The open reading frame (ORF) encoding RnhC (Rv2228c) was amplified from M. tuberculosis genomic DNA and cloned into pMAL-C2 (95) to produce an N-terminal maltose-binding protein (MBP) fusion construct which was modified to include a His₆ tag and a 3C protease recognition sequence (96) immediately upstream of the RnhC sequence. The plasmid was transformed into E. coli BL21(DE3)pLysS (97). Protein was produced by auto-induction of cultures grown in ZYP-5052 medium (98) for 7 h at 37°C and then for 20 h at 28°C. The cells were harvested (5,300 × g for 30 min at 4°C) and resuspended in lysis buffer (20 mM HEPES and 150 mM NaCl, pH 7.5) containing 10 μg/mL of DNase I and one cOmplete Mini EDTA-free protease inhibitor tablet (Roche) per 10 mL total volume. The cells were lysed using a cell disruptor (Constant Systems) at a pressure of 18 kilopounds per square inch (kpsi). The lysate was centrifuged at 20,000 × g for 30 min at 4°C. Recombinant RnhC was purified by affinity chromatography using an amylose column and cleaved from the tags (MBP and His₆) by overnight incubation of His₆-tagged recombinant 3C protease at 4°C. The 3C protease and the tags were removed by reverse affinity chromatography. RnhC was concentrated and further purified by size exclusion chromatography using a 16/600 Superdex 200 pg column (GE Healthcare). RnhC was eluted in a single peak, concentrated to 10 mg/mL, and stored at −20°C.

Inhibitors of HIV RNase H were selected from the PubChem database and kindly provided by the National Institutes of Health (NIH) through the Developmental Therapeutics Program (DTP). The compounds were resuspended in dimethyl sulfoxide (DMSO) at a final concentration of 10 mM and stored at −20°C.

The RNase HI activity assay was performed using a Förster resonance energy transfer (FRET) assay as previously described (99) using an 18-bp RNA/DNA substrate, with the RNA strand labeled with fluorescein at the 3′-end (5′-GAU-CUG-AGC-CUG-GGA-GCU-fluorescein-3′), and the cDNA strand labeled with Iowa black on the 5′-end (5′-Iowa black-AGC-TCC-CAG-GCT-CAG-ATC-3′). RnhC (4 nM) was incubated with 25 nM substrate in FRET buffer (50 mM Tris-HCl, pH 8, 60 mM KCl, 5 mM MgCl₂). The increase in fluorescein signal due to substrate hydrolysis was monitored at excitation/emission wavelengths of 490 nm/528 nm using a SpectraMax iD3 plate reader. Compounds active at an initial concentration of 100 μM were subsequently assessed using a dose-response assay.