Rapidly Correcting Frameshift Mutations in the Mycobacterium tuberculosis orn Gene Produce Reversible Ethambutol Resistance and Small-Colony-Variant Morphology

We have identified a previously unknown mechanism of reversible high-level ethambutol (EMB) resistance in Mycobacterium tuberculosis that is caused by a reversible frameshift mutation in the M. tuberculosis orn gene. A frameshift mutation in orn produces the small-colony-variant (SCV) phenotype, but this mutation does not change the MICs of any drug for wild-type M. tuberculosis. However, the same orn mutation in a low-level EMB-resistant double embB-aftA mutant (MIC = 8 μg/ml) produces an SCV with an EMB MIC of 32 μg/ml.

ABSTRACT

We have identified a previously unknown mechanism of reversible high-level ethambutol (EMB) resistance in Mycobacterium tuberculosis that is caused by a reversible frameshift mutation in the M. tuberculosis orn gene. A frameshift mutation in orn produces the small-colony-variant (SCV) phenotype, but this mutation does not change the MICs of any drug for wild-type M. tuberculosis. However, the same orn mutation in a low-level EMB-resistant double embB-aftA mutant (MIC = 8 μg/ml) produces an SCV with an EMB MIC of 32 μg/ml. Reversible resistance is indistinguishable from a drug-persistent phenotype, because further culture of these orn-embB-aftA SCV mutants results in rapid reversion of the orn frameshifts, reestablishing the correct orn open reading frame, returning the culture to normal colony size, and reversing the EMB MIC back to that (8 μg/ml) of the parental strain. Transcriptomic analysis of orn-embB-aftA mutants compared to wild-type M. tuberculosis identified a 27-fold relative increase in the expression of embC, which is a cellular target for EMB. Expression of embC in orn-embB-aftA mutants was also increased 5-fold compared to that in the parental embB-aftA mutant, whereas large-colony orn frameshift revertants of the orn-embB-aftA mutant had levels of embC expression similar to that of the parental embB-aftA strain. Reversible frameshift mutants may contribute to a reversible form of microbiological drug resistance in human tuberculosis.

INTRODUCTION

Tuberculosis (TB) remains a major health problem, with two million deaths and seven million new cases annually (1). The emergence of multidrug-resistant and extensively drug-resistant Mycobacterium tuberculosis poses a serious challenge to TB control programs in both developing and industrialized countries. An additional challenge to TB control efforts is the ability of M. tuberculosis bacilli to enter a tolerant state that enables them to be transiently drug resistant (2–4). An improved understanding of the molecular mechanisms that lead to drug resistance and tolerance may help efforts to control this disease.

Ethambutol (EMB) is a first-line antituberculosis drug that is used in the treatment of drug-susceptible and multidrug-resistant TB (5). Resistance to EMB has been associated with mutations in embCAB, ubiA, and aftA genes, and it seems to be likely to occur as a multistep process (6–8). The embCAB and aftA genes encode arabinosyl transferases involved in cell wall arabinogalactan and lipoarabinomannan biosynthesis (9–11). The decaprenylphosphoryl-5-phosphoribose (DPPR) synthase, UbiA, is involved in the synthesis of decaprenylphosphoryl-β-d-arabinose (DPA), which is the donor substrate for arabinosyl transferases (12, 13). Both epidemiological and laboratory studies suggest that high-level EMB resistance observed in embB mutant strains is stable and is associated with single or combined mutations in embC, embA, or ubiA (6, 7, 14). Until the present study, no mechanism of reversible high-level resistance to EMB has been described.

Recently, we described a new mechanism of drug tolerance that is caused by reversible and heritable genotypic switching of the M. tuberculosis glpK gene (15). Reversible frameshift mutations in a 7C homopolymeric sequence of glpK produce small-colony variants (SCVs) that have increased tolerance to antituberculosis drugs. Rapid reversion of these frameshift mutations reestablishes the glpK open reading frame and reverses both the SCV morphology and the drug-tolerant phenotype. Thus, this mechanism results in a type of reversible heritable phenotypic variation (15). Here, we show that reversible frameshift mutations in the M. tuberculosis orn gene can also produce SCVs, except that in this case, reversible resistance to only a single drug is produced, and the resulting MIC increase is substantially greater than that observed for drug-tolerant glpK mutants. This study suggests that reversible frameshift mutations are not confined to a single gene target but may represent a more widely used mechanism to adapt to stresses encountered by M. tuberculosis.

RESULTS

Reversible high-level resistance to ethambutol.

We had previously investigated the causes of an unstable drug resistance phenotype observed in the course of an in vitro study of resistance to the antituberculosis drug EMB (7). We passaged cultures of the drug-susceptible M. tuberculosis strain 210 (EMB MIC, 2 μg/ml) on progressively higher levels of EMB, selecting a series of related strains (4C31, 16C1, and 24C1) with stepwise increases in EMB MIC (Fig. 1). Whole-genome sequencing (WGS) showed that each new strain contained mutations that mapped to genes affecting the DPA biosynthetic and utilization pathway, which is related to the EMB mechanism of action (7), except for one highly EMB-resistant mutant, 24C1 (EMB MIC, 32 μg/ml). In the present study, further characterization of this highly EMB-resistant strain demonstrated that it had acquired a new single nucleotide frameshift insertion (C insertion at nucleotide 463) in the orn gene (Rv2511) compared to its precursor strain, 16C1 (EMB MIC, 8 μg/ml) (Fig. 1). Compared to strains 210, 4C31, and 16C1, strain 24C1 grew more slowly in liquid medium and produced small colonies on solid medium (Fig. 2A and B). Interestingly, colonies with normal size, equivalent to that of the parental strain 16C1, were detected at a frequency of 3.33 × 10⁻⁵ in subsequent cultures of 24C1 in liquid medium without added EMB. These revertant large colonies comprised 2% ± 1% and 70% ± 10% of the CFU after 4 and 11 serial passages, respectively (Fig. 2C and D). Furthermore, the large-colony 24C1 revertants also reverted their EMB resistance back to that of their 16C1 precursor, showing that both the 24C1 colony morphology and its high-level EMB resistance phenotype are unstable (Fig. 1). We then subjected strain 16C1 to the same selection conditions that produced 24C1 two additional times. Neither of these two attempts produced SCVs; instead, these additional selection studies generated only four large colonies in the combined experiments, each of which had orn genes with wild-type sequences.

FIG 1.

M. tuberculosis small-colony-variant orn frameshift mutants and ethambutol resistance. Stepwise evolution of M. tuberculosis strain 210 selected for progressively higher levels of ethambutol resistance in vitro. The ethambutol concentrations used for selection are shown at each step. The strains 210, 16C1, 24C1, and BC8 were whole-genome sequenced in this study. Spontaneous new mutations are shown in red, and new mutations introduced by allelic exchange techniques are shown in blue. Ethambutol MICs are shown in parenthesis. The asterisk indicates a genome that was previously sequenced using the Roche 454 platform and reported in a previous study.

FIG 2.

The orn genotype modulates growth phenotype in M. tuberculosis. (A) The different growth kinetics of the M. tuberculosis isogenic mutants in 7H9 medium are shown as optical density at 580 nm (OD₅₈₀). (B) Cultures were plated on 7H10 containing 10% oleic acid-albumin-dextrose-catalase (OADC) and 0.5% glycerol and incubated at 37°C for 4 weeks. The parent strain 210 and the revertant BC8, which have functional orn genes, showed normal-size colonies. Strains 24C1 and 210-ornInsC, with a frameshifted orn gene, displayed a small-colony phenotype. (C) Large colonies emerged after four passages of 24C1 in 7H9 medium not containing ethambutol. (D) After 11 passages, most of the 24C1 culture had reverted to large-colony morphology. In each panel, a representative result of three independent experiments is shown.

Second frameshift mutations restore the correct orn reading frame to revertant colonies.

A WGS analysis of one large-colony revertant, BC8 (Fig. 1), revealed the presence of a new orn mutation (deletion of G453) compared to 24C1 that reestablished the normal orn reading frame and produced a Lys153Arg change (Fig. 1). BC8 was also found to have an Ala7Thr mutation in mazF8 (Rv2274c). To further explore the genetic causes of SCV reversion to large colony size during serial passage of 24C1, we sequenced the orn gene in 171 additional large colonies randomly picked across three 24C1 reversion experiments (Fig. 3; also, see Table S1 in the supplemental material). Sanger sequencing revealed that all 171 of the revertants had one of eight distinct mutations in orn, all of which restored the correct orn reading frame through additional insertions or deletions (Fig. 3 and Table S1). Some mutants developed amino acid changes in the process of reestablishing the normal orn reading frame; however, interestingly, the most common mutation was a G462 deletion in a 4G repeat (present in 60% ± 7% of the revertants) that resulted in a synonymous mutation, Gly154Gly (GGG to GGC), and a reversion of the Orn protein sequence to that seen in the wild-type 16C1 strain (Fig. 3 and Table S1). We also sequenced mazF8 in 20 of these revertant colonies, and no mutations in this gene were found, suggesting that mazF8 was not involved in reversion. Thus, 24C1 along with its revertant BC8 illustrates a new type of rapidly reversible drug resistance that coincides with changing colony morphology and growth rates.

FIG 3.

Mutations in the orn gene of the small-colony strain 24C1 and large-colony revertants. Red arrows indicate the insertion of a C in the orn gene in 24C1, and each black arrow represents the second mutation identified in the revertant colonies. The number of colonies with the same orn genotype is given in parentheses. The resulting changes in the Orn protein are shown in boxes. Ins, insertion; Δ, deletion; *, development of a frameshift scar (C-ins plus an additional insertion or C-ins plus a deletion) in the orn gene of revertant colonies; **, development of a new single nucleotide polymorphism in the orn gene of a revertant colony.

We have named the type of frameshift mutations present in BC8 a “frameshift scar” (i.e., a frameshifting insertion or deletion situated near another insertion or deletion that reestablishes the frame). The presence of orn mutations in clinical M. tuberculosis isolates perhaps marks similar reversible frameshift events in clinical TB. An in silico analysis of 6,464 M. tuberculosis genomes deposited in the GenBank database (RefSeq genome database) revealed 6,198 genomes with a wild-type orn genotype, 188 with nonsynonymous mutations, 77 with synonymous mutations, and 1 with a deletion of 4 bp (Table S2). However, these mutations were not located in the same region as the in vitro mutations. Furthermore, the presence of possible genetic footprints of reversible frameshift mutations in the orn gene is only suggestive that such frameshifts have occurred in clinical strains, and confirmation will require additional clinical studies.

Functional consequences of orn mutations.

The M. tuberculosis orn gene encodes a versatile enzyme that can hydrolyze various substrates, such as nanoRNA (2 to 5 nucleotides long), 5′-p-nitrophenyl thymidylate (pNP-TMP) and 5′-O-phosphonoguanylyl-(3′→5′)-guanosine (pGpG) (16–21). We found that both the wild-type M. tuberculosis orn and the Lys153Arg mutant orn of the BC8 revertant complemented Escherichia coli strain UM341, which has a conditional orn deletion and produces an SCV phenotype (Fig. 4) (22). However, complementing E. coli UM341 with the M. tuberculosis orn C insertion (C-ins) frameshift mutant present in 24C1 failed to restore colony morphology to normal size (Fig. 4). In M. tuberculosis, replacing the wild-type orn in 210 or 16C1 with orn C-ins resulted in strains (210-ornInsC and 16C1-ornInsC, respectively) with an SCV phenotype (Fig. 2A and B and Table 1), although 210-ornInsC showed growth rates between those of the wild type and 16C1-ornInsC or 24C1 (Fig. 2A). Interestingly, while the SCV 16C1-ornInsC reverted to a large-colony phenotype after a mean of 8 serial passages, the SCV 210-ornInsC was stable for more than 20 passages (Table 1). Having confirmed that frameshift and frame reversion mutations in orn were responsible for the observed differences in colony size, we next investigated whether these mutations were also responsible for differences in EMB MIC (Table 1). Inserting orn C-ins into 16C1 did indeed increase the EMB MIC from 8 to 32 μg/ml, demonstrating that inactivating orn frameshift mutations in strain 16C1 produces high-level EMB resistance. However, 210 and 210-ornInsC had identical wild-type EMB MICs, demonstrating that the orn-inactivating mutations do not produce EMB resistance in a wild-type background. These results also show that production of an SCV may not be sufficient in itself to alter EMB MICs.

FIG 4.

Insertion of an in-frame but not a frameshifted M. tuberculosis orn gene complements an E. coli conditional orn deletion mutant. M. tuberculosis orn genes of the wild-type (WT-orn) 210 strain, the frameshift mutant (24C1-orn) 24C1 strain, and the revertant mutant (BC8-orn) BC8 strain were amplified and cloned into a pBAD18 vector. An E. coli strain with a conditional orn deletion was transformed with each plasmid, and the M. tuberculosis orn gene function was assessed by colony morphology, as described in reference 22. A representative result of three independent experiments is shown.

TABLE 1.

orn and aftA genotypes, embC expression, and ethambutol MICs in large- and small-colony variants of M. tuberculosis^a

Strain	orn gene	aftA gene	embC expression (fold change compared to 210)b	Colony morphology	EMB MIC (μg/ml)c	No. of passages to reversion
210	WT (active)	WT	1	Large	2	NA
210-ornInsC	Frameshift (inactive)	WT	1	Small	2	>20
16C1	WT (active)	Leu198Leu	5	Large	8	NA
16C1-ornInsC	Frameshift (inactive)	Leu198Leu	27	Small	32	8 ± 1
24C1	Frameshift (inactive)	Leu198Leu	27	Small	32	4 ± 1
BC8	MT (active)	Leu198Leu	8	Large	8	NA
16C1Δcas	WT (active)	Leu198Leu	ND	Large	8	NA
24C1Δcas	Frameshift (inactive)	Leu198Leu	ND	Small	32	6 ± 1

^aWT, wild-type; MT, mutant; EMB, ethambutol; NA, not applicable; ND, not done.

^bembC expression was determined by three independent RNA-seq experiments (Table S4).

^cAll MICs were determined in triplicate (Table S3). Because each value within a triplicate MIC test was similar to the other values, a single MIC is shown without standard deviation.

Inactivation of the orn gene further increases the level of embC expression induced by synonymous mutations in aftA.

We used RNA sequencing (RNA-seq) to study the transcriptional response of the wild-type 210, 16C1, 24C1, BC8, 210-ornInsC, and 16C1-ornInsC strains and determine whether the expression levels of genes involved in cell wall biosynthesis are altered by Orn inactivation (Fig. 5). RNA-seq analysis of the strains showed that levels of embC expression correlated with a strain’s EMB MIC (Table 1 and Table S4). We noted that strains 210 and 210-ornInsC had the lowest level of embC expression, while 16C1 and BC8 had intermediate embC expression (likely due to the presence of the AftA Leu198Leu mutation, which activates a cryptic promoter of embC [7]). Finally, both 24C1 and 16C1-ornInsC had the highest levels of embC expression, which appeared to be caused by Orn inactivation in the presence of an aftA mutant background. The levels of expression of embA, embB, and ubiA genes were not significantly altered. Overexpression of embC in M. tuberculosis and other mycobacteria elevates EMB MIC (8, 23), providing a plausible mechanism for the reversible increases in EMB MIC that we had observed in these strains.

FIG 5.

Two-way hierarchical clustered heat map of RNA-seq expression data. The heat map, generated using Pearson correlation, shows the abundance of transcript (rows) in each sample (column). Red represents high expression, blue represents low expression, and white represents median expression.

RNA-seq also showed high-level expression of the cas (CRISPR [clustered regularly interspaced short palindromic repeat]-associated) genes Rv2820c to Rv2824c (35- to 50-fold increase compared to strain 210) in 24C1 (Table S4), and these differences were confirmed by reverse transcription-PCR (RT-PCR) (Fig. S1). This high level of cas gene expression was also detected by RNA-seq profiling of the frameshifted-orn SCV strains 210-ornInsC and 16C1-ornInsC (Table S4). Conversely, the large-colony strains 16C1 and BC8 showed no change or a 3- to 8-fold increase of Rv2820c-Rv2824c expression compared to strain 210, respectively. This difference is perhaps due to differences in orn genotype: 16C1 is orn wild type and BC8 is an orn Lys153Arg mutant. The parental strain, 210, used in this study is a member of the M. tuberculosis Beijing clade, and members of this clade have a natural deletion of the cas genes Rv2816c to Rv2819c and part of Rv2820c (Fig. S2) (24, 25). We considered the possibility that modulation of the Rv2821c-Rv2824c operon seen in the embB-aftA-orn triple mutant might play a role in EMB resistance or even in reversion frequency. Thus, we constructed Rv2820c-Rv2824c deletions in both strains 16C1 and 24C1. This deletion did not affect the colony morphology, EMB MICs, or SCV reversion rates of either strain (Table 1).

DISCUSSION

Numerous clinical and laboratory studies have demonstrated that drug resistance in M. tuberculosis is caused by fixed mutations in drug-activating genes, genes encoding drug targets, the promoter regions of drug targets, and genes encoding drug efflux pumps (6, 7, 26–29). In contrast to drug resistance, drug tolerance has been associated with transiently heritable (15) and transiently nonheritable (2–4) mechanisms. Here, we demonstrate that even elevated levels of drug resistance can be caused by a transiently heritable mutation. Resembling some forms of drug tolerance, this transiently heritable drug resistance is caused by a reversible frameshift mutation in a target that is not directly associated with drug resistance. The mutation we observed is also associated with a shift to a small-colony phenotype. To our knowledge, this is the first description of an association between two-step frameshift corrections and reversible high-level drug resistance in M. tuberculosis.

We showed previously that mutations in embC and very high embC expression levels are associated with high-level EMB resistance (7). The current study expands the potential causes of high-level EMB resistance to include frameshifting mutations in the M. tuberculosis orn gene. Inactivation of orn increases embC expression to a very high level in an aftA mutant background but not in the presence of wild-type aftA; thus, synonymous aftA mutations appear to be required for orn-mediated high-level EMB resistance. EmbC is one of the cellular targets of the drug EMB (8). Overexpression of embC in vitro causes EMB resistance (8, 23), and laboratory-evolved highly EMB-resistant M. tuberculosis mutants are also associated with the development of new embC mutations (7). Orn is a versatile enzyme that can hydrolyze various substrates. We have found that inactivation of orn induces high levels of cas gene (Rv2821c-Rv2824c) expression. Our strains are derived from a W-Beijing isolate with a natural deletion of part of the cas operon (ΔRv2816c-Rv2820c). Therefore, the role of the complete cas operon in orn SCV and EMB phenotype remains to be investigated in other M. tuberculosis lineages. M. tuberculosis contains a gene for a second Orn-like exoribonuclease, Rv2837c, that also degrades nanoRNAs (22) and is involved in cyclic-di-AMP and cyclic-di-GMP hydrolysis (30–32). Also similar to our orn frameshift mutants, deletion of Rv2837c increased expression of the cas genes, which are mediated by changes in nanoRNA levels rather than by elevated cyclic-di-nucleotide (cyclic-di-NMP) levels (33). This provides indirect evidence that orn inactivation also elevates levels of small nucleotides and nanoRNAs. However, the mechanism by which bioactive nucleotides might increase expression of a cryptic embC promoter remains to be elucidated.

Our results show that SCV orn mutants revert to a large-colony morphology and more EMB-susceptible phenotype when cultured in the absence of EMB. We noted that almost 60% of the orn mutants that reverted to the original phenotype did so by acquiring a synonymous Gly154Gly (G-to-C) mutation. Based on a comparison to the parental 24C1 strain, we conclude that this synonymous mutation actually occurred as a result of two events: first, the insertion of a C (to cause the frameshift), and second, the deletion of a G (to reestablish the frame). This chain of events would have been difficult to detect in clinical isolates and instead would have been assumed to be the result of a single nucleotide polymorphism in the gene. Thus, it is possible that reversible frameshift mutations causing transient drug resistance might be confused clinically with phenotypically reversible drug persistence in human TB cases.

Transiently heritable drug tolerance in M. tuberculosis has been strongly associated with frameshift mutations in a 7C homopolymeric tract located within the M. tuberculosis glpK gene. Reversible frameshift mutations in the glpK 7C tract occur at high frequency (10⁻²), which suggests slipped-strand mispairing associated with mismatch repair deficiency as the mutational mechanism (15). Similarly, the reversible orn frameshift mutations occurred at a frequency of 10⁻⁵, with the majority detected in a tetranucleotide 4G sequence located close to the original orn frameshift mutation responsible for the SCV phenotype. These frequencies are very high compared to those of other bacteria with functional mismatch repair genes (10⁻⁹ to 10⁻⁷) (34). Taken together, these results suggest that frameshift mutations occur at high frequencies in homopolymeric and nonhomopolymeric regions of the M. tuberculosis genome, similar to other mismatch repair-deficient mycobacteria (35, 36). These findings suggest that reversible frameshift mutations may also play a role in regulating other cellular functions in M. tuberculosis, such as stress responses, by toggling contingency gene expression on and off. In fact, reversible frameshift mutations may provide a means of flexible adaptive mechanisms to different types of cell stress that could be widespread in the genome of M. tuberculosis. Further investigations are needed to identify the potential participation of other reversible frameshift mutations in M. tuberculosis stress adaptation and their contribution to TB drug resistance and pathogenesis.

MATERIALS AND METHODS

Bacterial culture and MIC testing.

M. tuberculosis strain 210 is a pansusceptible W-Beijing isolate obtained from Kathleen Eisenach at the University of Arkansas for Medical Sciences (37, 38). Strain 210 and its isogenic strains were cultivated at 37°C either in Middlebrook 7H9 broth (Difco) containing 0.05% Tween 80 or on Middlebrook 7H10 agar supplemented with 0.5% glycerol, both enriched with 10% oleic acid-albumin-dextrose-catalase (Difco). Broth cultures were incubated with gentle shaking. The MICs of EMB, isoniazid (INH), and rifampin (RIF) were determined by the 7H10 agar proportion method, as described previously (27, 39), and were confirmed by testing in triplicate. For all plasmid construction, Escherichia coli strain Top10 (Invitrogen) was grown in Luria-Bertani broth or agar (both from Sigma-Aldrich) at 37°C, supplemented with 50 μg/ml kanamycin (Sigma-Aldrich) and/or X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; Invitrogen), where appropriate.

DNA isolation, PCR, and bidirectional DNA sequencing.

Genomic DNA was extracted using the cetyltrimethylammonium bromide (CTAB) protocol (40, 41). One loop of M. tuberculosis agar culture was suspended in 1× Tris-EDTA buffer and heated at 80°C for 1 h. The heat-killed bacteria were lysed using lysozyme incubation overnight followed by proteinase K and sodium dodecyl sulfate (SDS) treatment. The aqueous phase containing nucleic acids was obtained by CTAB extraction followed by chloroform-isoamyl alcohol purification. The aqueous phase was treated with RNase I, and the DNA was precipitated by ethanol. To amplify the orn gene for DNA sequencing, PCR was performed using a mix containing 5 ng of genomic DNA, 5 pmol of each primer (F1-primer, 5′TGCCCATAGCTGGATCACAC3′; R1-primer, 5′TCACAACCTGGTGCTCTACC3′), 200 μM deoxynucleoside triphosphates (dNTPs), 1× PCR buffer, 1× GC enhancer, and 2 U of Platinum SuperFi DNA polymerase (Invitrogen) per 50-μl reaction mixture. All PCR products were examined on an ethidium bromide-stained agarose gel and purified using a gel extraction kit (Qiagen). Direct Sanger sequencing of PCR products was performed with a BigDye Terminator kit and analyzed with an ABI3100 genetic analyzer (Applied Biosystems).

In vitro selection and isogenic strain construction.

The strains 4C31, 16C1, and 24C1 were previously isolated by multistep selection for ethambutol resistance (7). In the first step, strain 210 was grown in 7H9 liquid medium without EMB to stationary phase (optical density at 580 nm [OD₅₈₀] = 1.6), and a dilution of approximately 1 × 10⁶ CFU was plated on 7H10 agar medium containing 4 μg/ml EMB and incubated for 4 weeks at 37°C. One colony was randomly picked and named 4C31. In the second step, 4C31 was grown in 7H9 liquid medium without EMB to stationary phase, and approximately 1 × 10⁹ CFU were plated on 7H10 agar medium containing 16 μg/ml EMB and incubated for 4 weeks at 37°C. One colony was randomly picked and named 16C1. In the third step, 16C1 was grown in 7H9 liquid medium without EMB to stationary phase, and approximately 1 × 10⁹ CFU were plated on 7H10 agar medium containing 24 μg/ml EMB and incubated for 4 weeks at 37°C. One colony was picked and named 24C1. To introduce a C insertion in the orn gene of M. tuberculosis strains 210 and 16C1, a 2,089-bp fragment spanning the C insertion was amplified from 24C1 genomic DNA using F2-primer (5′GGGAAGCTTGTGTAGATTCCTGGGTAGCG3′) and R2-primer (5′AAAGGTACCACAACACGACGAATGGATCG3′). The PCR product was purified, digested with HindIII and Acc65I, and cloned into the p2NIL vector, followed by the insertion of the PacI cassette containing the sacB and lacZ genes (26, 27, 42). The recombinant plasmids were used to transform M. tuberculosis strains, mutants were generated, and blue colonies (single crossovers) were grown on 7H10 agar medium without selection and plated on sucrose–X-Gal plates to select for white mutant colonies (double crossovers) (26, 27, 42). The mutant colonies were screened for C insertion in the orn gene by PCR and Sanger sequencing as described above.

Reversion of small-colony variants to parent large-colony phenotype.

The number of culture passages needed for 24C1, 16C1-ornIns, and 210-ornInsC strains to revert to the large-colony phenotype was determined as previously described (15). Briefly, three independent colonies from 7H10 agar plates were randomly picked and grown in 7H9 liquid medium to stationary phase (OD₅₈₀ = 1.6), which was considered passage 0. The cultures were diluted to 10⁶ to 5 × 10⁶ CFU per ml and incubated at 37°C for 1 week, which was considered passage 1. The passages were repeated up to 20 times, and each culture was plated on 7H10 agar medium to determine the number of large colonies. To calculate reversion frequencies, cultures were plated at high density to determine revertant count and at low density to determine total count. The results were expressed as the mean number of revertants obtained divided by the number of total CFU.

Whole-genome sequencing and mutation detection.

Total genomic DNA was extracted as described above and purified by using a MagAttract high-molecular-weight (HMW) DNA kit (Qiagen). The DNA of 210, 16C1, BC8, and 24C1 was submitted for whole-genome sequencing to The Institute for Genome Sciences, University of Maryland School of Medicine, Baltimore, MD, USA. Briefly, Genomic DNA libraries were constructed for sequencing on the Illumina platform using the KAPA library preparation kit (Kapa Biosystems, Woburn, MA). First, DNA was fragmented with the Covaris E210 instrument. Then, libraries were prepared using a modified version of the manufacturer’s protocol. The DNA was purified between enzymatic reactions, and the size selection of the library was performed with AMPure XP beads (Beckman Coulter Genomics, Danvers, MA). The PCR amplification step was performed with primers containing an index sequence six nucleotides in length. Libraries were sequenced on an Illumina HiSeq2000 sequencer with a 101-bp paired-end run. The reads were aligned to M. tuberculosis H37Rv (GenBank ID AL123456.3), and single nucleotide polymorphisms and indels (insertions and deletions) were detected, quality filtered, and annotated.

RNA extraction, library preparation, and sequencing.

M. tuberculosis strain 210 and isogenic strains 24C1, 16C1, BC8, 210-ornInsC, and 16C1-ornInsC were grown to mid-log phase (OD₅₈₀ = 0.6 to 0.7) at 37°C in 7H9 medium. Total RNA was extracted by using TRIzol reagents (Invitrogen) as described previously (43). The extracted RNAs were submitted for transcriptome analysis to The Genomics Center, Rutgers University, Newark, NJ, USA. RNA library construction and sequencing were performed as described previously (15). For each strain, three different RNA preparations from three independent cultures were used for RNA sequencing (RNA-seq) analysis. The statistical analysis of differentially expressed genes compared to 210 was performed using CLC Genomic Workbench as described previously (15). The differences in mRNA expression were assessed by false-discovery-rate (FDR) analysis with an adjusted P value of ≤0.05.

Statistical analysis.

Data analysis was performed using a statistical tool of CLC Genomic Workbench or GraphPad Prism 8.4.1, as appropriate. The CLC tool was used to analyze RNA-seq data, and differentially expressed mRNAs were defined using a false-discovery-rate-adjusted P of ≤0.05. GraphPad Prism 8.4.1 was used to perform one-way analysis of variance (ANOVA) of mutation frequency detected in revertant large-colony strains, and statistical significance (P value of ≤0.05) was determined using the Holm-Sidak method.

Data availability.

Whole-genome sequencing (WGS) and RNA-seq raw data were deposited in the NCBI BioProject database (ID PRJNA478476). The sample accession numbers for genomes and transcriptomes analyzed in this study are described in Table S5.