Characterization of Large Deletion Mutants of Mycobacterium tuberculosis Selected for Isoniazid Resistance

INTRODUCTION

We previously reported the construction of isoniazid (INH)- and rifampin (RIF)-resistant Mycobacterium tuberculosis strains that may be used safely in biosafety level 2 laboratories (1). These multidrug-resistant strains were based on M. tuberculosis strains auxotrophic for the nutrients pantothenate, leucine, and either methionine or arginine. Surprisingly, selection for INH resistance in the pantothenate/leucine/arginine auxotroph led to the isolation of two multidrug-resistant strains, mc²8248 and mc²8250 (1), with large genomic deletions (LGDs) (46 to 48 kbp) encompassing katG (Rv1908c), encoding a catalase peroxidase, the activator of INH (2). All INH-resistant mutants isolated from the pantothenate/leucine/methionine auxotroph had only point mutations in katG, and no deletion encompassing katG, small or large, was found (1). We therefore sought to examine whether the arginine auxotrophy was specifically responsible for the formation of these LGDs during INH resistance selection.

Nonmutagenized cultures of wild-type H37Rv and two arginine-deficient mutants, H37Rv ΔargB and H37Rv ΔargF (3), were grown at 37°C in Middlebrook 7H9 medium supplemented with OADC (oleic acid-albumin-dextrose-catalase-sodium chloride)-glycerol-tyloxapol and nutrients (1, 4). The cultures were plated on Middlebrook 7H10-OADC-glycerol plates containing INH (7.3 μM) and nutrient supplement. Isolated colonies were picked and patched onto plates with or without INH to confirm INH resistance. The INH-resistant mutants were then screened for the presence of katG by PCR. Whereas katG was present in the 13 screened INH-resistant mutants isolated from H37Rv, katG deletion, partial or full, was detected in 17 of 44 and 2 of 12 INH-resistant mutants isolated from H37Rv ΔargB and H37Rv ΔargF, respectively. Mutants with a potential katG deletion were then subjected to whole-genome sequencing (WGS). WGS was performed on a MiSeq instrument (Illumina, CA, USA). Libraries were prepared, normalized, and pooled using the Nextera XT library kit following the manufacturer’s instructions. Coverage depth ranged from 18- to 60-fold. All short reads (75 bp) were mapped against the H37Rv reference genome. The regions with zero coverage were determined to be deletions. Nine unique LGDs ranging from 14 to 64 kbp were identified (Fig. 1). All of the LGDs obtained were resistant to INH (MIC, >30 μM) (Table 1).

• Open in new tab
• Download powerpoint

FIG 1

Deleted regions in the INH-resistant mutants. The LGD regions are compared with the H37Rv genomic region between Rv1857 and Rv1932. Dotted lines demark the deleted region in each mutant. The beginning and end coordinates for each deletion are indicated below the dotted lines. The position of katG is delineated by two vertical lines.

LGDs encompassing katG were previously identified in INH-resistant M. tuberculosis clinical isolates. Small (2.3-kbp) to large (12.0- and 34.4-kbp) deletions were found in INH-resistant M. tuberculosis strains isolated from Japanese patients (5). Seven LGDs (3 to 53 kbp) were identified in multidrug-resistant and extensively drug-resistant M. tuberculosis clinical strains belonging specifically to East Asian lineage 2 (6). Additionally, a 19.5-kbp deletion was found in a clinical multidrug-resistant M. tuberculosis Beijing strain (7). These studies highlighted a possible link between LGD formations in INH-resistant strains and M. tuberculosis strains from East Asian lineage. This led us to isolate INH-resistant mutants in vitro from a nonmutagenized culture of M. tuberculosis strain Beijing HN878. katG from 2 of 12 mutants could not be amplified by PCR, and WGS revealed the presence of small (5.7-kbp) and large (37.5-kbp) genomic deletions in these 2 mutants (Fig. 1). This established that the isolation of LGDs in INH-resistant M. tuberculosis strains did not depend on strains harboring mutations in the arginine biosynthesis pathway. Furthermore, an LGD (49 kbp) encompassing katG had been identified in an INH-resistant mutant isolated in vitro from an RIF-resistant M. tuberculosis Erdman strain (8). Our laboratory strain collection contained one INH-resistant mutant isolated from an RIF-resistant H37Rv strain from which no katG PCR product could be obtained. We therefore performed WGS on that multidrug-resistant strain, which revealed a 57-kbp LGD (mc²5000) (Fig. 1).

Focusing on the LGDs derived from M. tuberculosis H37Rv, we tested their growth at 37°C (Fig. 2). The growth kinetics were similar in all the mutants compared to their parental strains, except for the mutant with the largest deletion (mc²5893). Because all of these LGDs lacked katG, we tested their growth in medium supplemented with ADS (albumin, sodium chloride, and dextrose) instead of OADC (no catalase present in the ADS medium). Again, only mc²5893 showed a growth defect under this condition (Fig. 2). We then tested their sensitivity toward oxidative stress by determining the MIC of paraquat (superoxide anion generator), hydrogen peroxide, and mitomycin C (double-stranded DNA damage) as described previously (Table 1) (9). The strains showed no substantial change in sensitivity to mitomycin C but increased sensitivity to paraquat and hydrogen peroxide correlating with a larger deletion.

• Open in new tab
• Download powerpoint

FIG 2

Growth of LGD mutants in Middlebrook 7H9 supplemented with glycerol, tyloxapol, and OADC (A) or ADS (B). M. tuberculosis cultures were grown to log phase (optical density at 600 nm [OD₆₀₀], 0.7 to 1.0), centrifuged, washed once with phosphate-buffered saline, and diluted 1/500 in the appropriate medium. Growth was followed by measuring OD_600nm three times a week. When required, arginine (1 mM) was added to the growth medium. Graphs show single replicates representative of at least two independent experiments.

To examine whether these strains with LGDs could infect and grow in macrophages, where M. tuberculosis isolates reside in vivo, growth in the murine macrophage cell line J774 was assessed as previously described (10). To avoid the effects that the addition of a large amount of arginine may have on regulating macrophage activation (11–13), only mc²5000 (H37Rv rpoB H445R Δ′Rv1874-′Rv1929c) was tested and compared with its parental strain mc²4986, H37Rv ΔkatG, and a multidrug-resistant M. tuberculosis strain with a point mutation in katG (mc²5857) (Fig. 3). A replication defect was observed in mc²5000 that could be attributed to its LGD because H37Rv ΔkatG and mc²5857 replicated during the first 24 h of infection before losing viability.

• Open in new tab
• Download powerpoint

FIG 3

Viability of the LGD mc²5000 (H37Rv rpoB H445R Δ′Rv1874-′Rv1929c) in J774 mouse cell line. J774 macrophage-like cells were infected with mc²4986 (parental strain of mc²5000, H37Rv rpoB H445R), mc²5000 (LGD), mc²5857 (H37Rv rpoB S450L katG L546P), and H37Rv ΔkatG at a multiplicity of infection of 4. At the indicated time points, J774 cells were lysed, and the lysates were plated to determine bacterial CFU. Graph represents the average of two independent experiments done in duplicate, shown as means with standard deviations (error bar).

In summary, 12 novel LGDs encompassing katG have been isolated on INH resistance selection in M. tuberculosis strains. Although we initially thought that a defect in arginine biosynthesis was responsible for the incidence of these LGDs, further work demonstrated that the LGDs could be obtained in M. tuberculosis strains fully functional for arginine biosynthesis.

Previously, LGD formation was postulated to happen through homologous recombination (5); however, we did not identify any potential site of recombination in our LGDs. The LGDs isolated in this study represent up to 76 kbp (1.7%) of the M. tuberculosis genome deleted. When combining all the LGDs, a total of 74 genes from Rv1858 to Rv1931c were removed on INH resistance selection (Table 2). Most of these genes have no known function, but they are not essential for M. tuberculosis growth in vitro. However, genes in this region may play a role in sensitivity to oxidative stress and growth in the intracellular environment. Further studies are needed to decipher whether a specific gene or a combination of genes included in these LGDs is responsible for these phenotypes.