Mutagenesis in the α3α4 GyrA Helix and in the Toprim Domain of GyrB Refines the Contribution of Mycobacterium tuberculosis DNA Gyrase to Intrinsic Resistance to Quinolones

Abstract

The replacement of M74 in GyrA, A83 in GyrA, and R447 in GyrB of Mycobacterium tuberculosis gyrase by their Escherichia coli homologs resulted in active enzymes as quinolone susceptible as the E. coli gyrase. This demonstrates that the primary structure of gyrase determines intrinsic quinolone resistance and was supported by a three-dimensional model of N-terminal GyrA.

Mycobacterium tuberculosis is intrinsically resistant to the majority of available antibiotics, including β-lactams, macrolides, and tetracyclines (20). Previous studies have investigated mechanisms of intrinsic resistance, including low permeability of the mycobacterial cell wall (16), production of inactivating enzymes such as β-lactamases (26), and constitutive efflux pumps (9). M. tuberculosis also exhibits a low level of intrinsic quinolone resistance, with MICs 20-fold (levofloxacin, sparfloxacin) to 140-fold (ciprofloxacin) higher than those for Escherichia coli (10).

Quinolones target the prokaryotic type II topoisomerases, DNA gyrase, and topoisomerase IV (7). Since M. tuberculosis does not possess a topoisomerase IV, DNA gyrase is the key quinolone target. The site of this interaction is likely the quinolone resistance-determining regions (QRDRs), classically domains 67 to 106 in GyrA and 426 to 464 in GyrB, where the majority of mutations conferring quinolone resistance are found (14, 25).

We hypothesized that the primary structure of DNA gyrase was mainly responsible for the intrinsic quinolone resistance of M. tuberculosis, since quinolone inhibition is 10-fold lower for mycobacterial gyrase than for E. coli gyrase (2, 10, 11). By comparing the quinolone MIC, QRDR sequence, and gyrase inhibition of M. tuberculosis to those of other mycobacterial species and E. coli, we previously suggested that an alanine residue at E. coli position 83 in GyrA (A83) is a determinant of resistance (2, 10). In the present study, we replaced A83 and other residues in the M. tuberculosis GyrA (α3α4 helix) and GyrB (topoisomerase-primase [Toprim] domain) (1) QRDRs with those present in E. coli. We found that the A83S substitution in GyrA, in synergy with M74I in GyrA and R447K in GyrB, confers quinolone susceptibility similar to that of E. coli gyrase. This study revealed that the structure of the antibiotic target is a mechanism of the intrinsic resistance of M. tuberculosis.

First, M. tuberculosis was compared to 21 bacterial species from five phyla, particularly intrinsic quinolone susceptibility and QRDR sequences of GyrA and GyrB (Fig. 1). Sequences were obtained from the SWISS-PROT/TrEMBL (http://expasy.org/sprot) and NCBI (http://www.ncbi.nlm.nih.gov) data banks and aligned using PipeAlign (22). QRDR similarities between species ranged from 47.5% to 100% for GyrA and from 54% to 97.5% for GyrB. Amino acids conserved between the species consisted of nine positions in GyrA (positions 75, 78, 79, 80, 82, 86, 90, 91, and 94) and thirteen in GyrB (positions 427 to 430, 435, 441, 442, 444, 448, 449, and 452 to 454). Apart from the alanine residue at position 83 in GyrA, we sought residues that differentiated M. tuberculosis and E. coli and varied according to quinolone susceptibility. Residues were conserved within individual phyla or among several phyla at the positions with amino acid variations. For example, I74 in GyrA and K447 and S464 in GyrB characterized proteobacteria, whereas M- or L74, W96, and P101 characterized actinobacteria, including mycobacteria. We chose six amino acids in GyrA (M74, A83, S84, I85, W96, and P101) and two in GyrB (R447 and N464) as candidates in the intrinsic quinolone resistance of M. tuberculosis.

FIG. 1.

Amino acid sequence alignment of the QRDRs in the A (GyrA) and B (GyrB) subunits of DNA gyrase for 21 bacterial species. Dashes represent amino acids identical to those in E. coli. Quinolone susceptibility is indicated by the ofloxacin MIC₅₀ (μg/ml) and the IC₅₀ (μg/ml) values for DNA supercoiling when available. (A) QRDR in GyrA from positions 67 to 106. (B) QRDR in GyrB from positions 426 to 464 (numbering system of E. coli). nd, not determined.

Expression plasmids encoding GyrA (A^A83S, A^M74I, A^M74I+A83S, A^{A83S+S84A+I85V}, A^{A83S+S84A+I85V+W96F+P101M} [called A^α4], and A^{S69V+A71G+E72D+T73V+M74I+N76K+A83S+S84A+I85V+W96F+P101M} [called A^QRDR]) and GyrB (B^R447K and B^N464S) subunits of M. tuberculosis were obtained from plasmids pATB and pBTB, in which mutations were introduced (2) using a QuickChange site-directed mutagenesis kit (Stratagene) and primers (MWG, Ebersberg, Germany) described in Table 1. Overexpression and subsequent purification of wild-type (wt) and modified GyrA and GyrB proteins were carried out as described previously (3, 17). A quantity of 0.2 to 1 mg/ml of each protein was obtained in the soluble fraction of the cell extract with a purity of >90% (see Fig. S1 in the supplemental material). The effect of quinolone was measured by two previously described topoisomerase assays (2, 3): the quinolone inhibition of DNA gyrase supercoiling (50% inhibitory concentrations [IC₅₀s]) (Fig. 2) and the induction of double-stranded DNA cleavage (concentrations that caused 25% linearization of the input DNA [CC₂₅s]) (see Fig. S2 and S3 in the supplemental material). Variability between experiments was minimized by performing the wt and mutant assays in parallel on the same day and under identical conditions. Enzyme assays were repeated two to four times, with the main results, such as those for A83S, being repeated most. Representative values are presented in Table 2. A combination of wt and mutant GyrA or GyrB subunits produced active DNA gyrase with specific activities close to that of the wt (Table 2), except for A^QRDRB^wt, which had no observed supercoiling activity, and A^wtB^R447K, which had very low cleavage activity.

TABLE 1.

Primers used in site-directed mutagenesis of Mycobacterium tuberculosis gyrA and gyrB genes^a

Mutation(s) introducedb	Nucleotide sequence (5′-3′)c	Matrix gene (nucleotides, 5′-3′)
A83S	CACCCGCACGGCGACTCGTCGATCTACGACAGCCTG	gyrA wt (253-287)
M74I	CGGTCGGTTGCCGAGACCATCGGCAACTACCACCCG	gyrA wt and A83S (223-258)
A83S S84A I85V	CACCCGCACGGCGACTCGGCGGTCTACGACAGCCTGGTG	gyrA A83S (253-290)
F96W P101M	GCCCAGCCCTTCTCGCTGCGCTACATGCTGGTGGAC	gyrA A83S S84A I85V (297-333)
S69V+A71G+E72D+T73V+M74I+N76K
Step 1	TCGGCCCGTTGGTTGCCGAGGCCATCGGCAAATACCACCCG	gyrA Aα4 (217-258)
Step 2	TCGGCCCGGGTGGTTGCCGACGCCATCGGCAAATACCACCCG	gyrA Aα4 (217-258)
Step 3	TCGGCCCGGGTGGTTGGCGACGTCATCGGCAAATACCACCCG	gyrA Aα4 (217-258)
R447K	GCGATACTTCCGCTGAAGGGCAAGATCATCAAT	gyrB wt (1546-1578)
N464S	GACCGGGTGCTAAAGAGCACCGAAGTTCAGGCGAT	gyrB wt (1597-1631)

^aOnly sense primers are presented; antisense primers are reverse and complementary.

^bThe first letter is the wt residue found in the M. tuberculosis gyrase sequence, the second letter is the substitution result, and the number corresponds to the codon position according to the E. coli numbering.

^cMutated nucleotides are in bold characters.

FIG. 2.

DNA supercoiling assay comparing ofloxacin inhibition of the wt M. tuberculosis DNA gyrase (AB) and the gyrase constituted with altered subunits (A^A83SB, A^M74IB, and A^A83S+M74IB). Relaxed pBR322 DNA (TR) was incubated with DNA gyrase in the absence (0) or presence of the indicated amounts of ofloxacin (μg/ml). R and S denote relaxed and supercoiled DNA, respectively.

TABLE 2.

Specific activities (in U/mg),^a inhibitory concentrations of quinolones for supercoiling (IC₅₀s in μM), and quinolone concentrations inducing cleavable complex (CC₂₅s in μM) for the M. tuberculosis gyrases reconstituted with wt or modified GyrA and GyrB proteins

Mutation(s) in:		Supercoiling sp act	IC50b				CC25
GyrAc	GyrB		MXF	OFX	ENX	NAL	MXF	OFX
wt	wt	1 × 104	6	28	156	4,750	2	7
A83S	wt	4 × 104	2.5	7	16	950	0.5	1.5
A83S S84A I85V	wt	2 × 103	2.5	8	11	2,800	1	3
α4	wt	2 × 103	2.5	7	16	1,080	1	3
M74I	wt	3 × 103	6	25	>160	3,450	2.5	ND
M74I A83S	wt	9 × 103	1	3	9	850	0.5	1.5
QRDR	wt	NO	NO	NO	NO	NO	0.5	0.5
wt	R447K	5 × 103	1	7	23	1,080	NO	NO
wt	N464S	2 × 103	7	20	ND	4,100	3.5	14
A83S	R447K	5 × 103	2.5	4	19	850	NO	NO
M74I A83S	R447K	5 × 103	2	11	16	520	NO	NO
GyrA of E. coli	GyrB of E. coli	1 × 106	2.5	1.5	14	215	0.5	0.2

^aOne unit is the amount of protein needed to produce 50% supercoiling of 0.4 μg of relaxed pBR322 DNA in 1 h at 37°C.

^bMXF, moxifloxacin; OFX, ofloxacin; ENX, enoxacin; NAL, nalidixic acid; NO, not observed; ND, not determined.

^cα4, substitutions A83S S84A I85V F96W P101M; QRDR, substitutions S69V A71G E72D T73V M74I N76K A83S S84A I85V W96F P101M.

Impact of gyrase substitutions on the quinolone effect.

In an assay testing supercoiling in the presence of moxifloxacin or enoxacin, the A83S substitution in GyrA was sufficient for obtaining the E. coli DNA gyrase inhibition values (Table 2). In contrast, to reach ofloxacin inhibitory values comparable to those of E. coli DNA gyrase, the A83S substitution had to be coupled with a second substitution, either M74I in GyrA or R447K in GyrB (Table 2). This shows that quinolones interact differently with gyrases depending on their structure and emphasizes the need for 3-D resolution of the gyrase quinolone pocket.

In the DNA cleavage assay, even if the A83S substitution was sufficient to obtain a moxifloxacin CC₂₅ value equal to that of E. coli DNA gyrase, none of the substitutions obtained this result for ofloxacin, except A^QRDR. Ofloxacin CC₂₅ values for E. coli DNA gyrase were still sevenfold lower than the CC₂₅s obtained for the A^A83SB and A^M74I+A83SB mutants. The IC₅₀ of nalidixic acid did not equal that of E. coli DNA gyrase for any of the mutants. However, DNA cleavage promoted by enoxacin and nalidixic acid was dramatically enhanced in the presence of the A83S substitution, compared to no cleavage by the wt M. tuberculosis DNA gyrase with nalidixic acid and weak cleavage with enoxacin (see Fig. S3 in the supplemental material). This is consistent with the previous observation that DNA cleavage was abolished in the S83F mutant of Streptococcus pneumoniae DNA gyrase (21) and highlights the important role of serine 83 in the mechanism underlying quinolone entrapment by the enzyme.

In addition to amino acid 83, the α4 helix has been shown to be involved in the intrinsic quinolone resistance of the Staphylococcus aureus DNA gyrase (24). We eliminated a possible role of amino acids 96 and 101 in M. tuberculosis GyrA, which are located deep in the core of the structure and did not seem to interfere with DNA or quinolones in our model described below (Table 2). Therefore, we found that in the α4 helix, only A83, located at the beginning of the α4 helix, played a role in intrinsic resistance and that modification of its vicinity (residues 84 and 85) did not have any effect toward quinolone susceptibility. This result was concordant with previous reports of a decrease in the quinolone affinity of E. coli DNA gyrase harboring S83A (12, 28). The key role of amino acid 83 in GyrA is also supported by the preponderance of amino acid substitutions found at position 83 as they relate to the acquired quinolone resistance of a large number of common bacterial species (14) as well as mycobacteria (3, 5, 23, 25). Alignment of GyrA QRDR sequences revealed an alanine at position 83 not only in mycobacteria but also in Ehrlichia (18), some soil bacteria (27), and Treponema pallidum, whereas there is a serine or threonine 83 in most other species (proteobacteria, firmicutes, and bacteroidetes), and rarely another residue in Borrelia burgdorferi (Q83) and Helicobacter pylori (N83) (Fig. 1A). Strikingly, some H. pylori strains have a threonine at position 83 and are more susceptible to quinolones than those with N83 (6), confirming that S83 or T83 confers quinolone susceptibility.

Comparison of the GyrA QRDR outside the α4 helix found that residue 74, lying in the α3 helix, varies according to intrinsic quinolone susceptibility. Residue 74 is an isoleucine in proteobacteria (except in Ehrlichia chaffeensis, but it is not quinolone susceptible), the phylum most susceptible to quinolones, and a methionine in other phyla, such as bacteroidetes, spirochetes, firmicutes, and actinobacteria like M. tuberculosis, characterized by lesser susceptibility to quinolones than proteobacteria (Fig. 1A). Our results showed that even if the M74I substitution alone had no effect on quinolone susceptibility, its association with A83S resulted in a more-susceptible enzyme (Table 2).

Molecular modeling of the breakage-reunion domain of M. tuberculosis DNA gyrase.

In order to integrate our results with the three-dimensional (3-D) structure, we attempted to model the GyrA α3α4 helix of M. tuberculosis DNA gyrase (Fig. 3) by homologous modeling of the E. coli GyrA 59-kDa backbone (PDB code 1AB4) (19) using the computer program Swiss-PDB viewer v3.7. The energy-minimized structure was obtained in 100 steps with a dielectric constant of 1, using the program CNS (4). The first interesting result from the model was that the GyrA QRDR is in close proximity to the catalytic tyrosine with a distance of less than 15 Å with residue 83 (Fig. 3). It is probable that residue 83, since it is located on the surface of the α4 helix, directly interacts with DNA. Therefore, S83 in E. coli could lead to the formation of a hydrogen bond between the gyrase and DNA, stabilizing the interaction between the two components, whereas A83 in M. tuberculosis cannot. In the M. tuberculosis configuration, interaction of M74 with M74 of the second monomer (less than 3 Å apart) may hinder the positioning of DNA in the DNA gyrase pocket and consequently lead to a weaker interaction between DNA and gyrase. This may be consistent with the previous observation that M. tuberculosis DNA gyrase has lower supercoiling activity than that of E. coli (specific activities of 10⁴ U/mg and 10⁶ U/mg, respectively) (2). As quinolones act on the gyrase-DNA binary complex, weaker complex stability could explain, in part, the weak stability of the ternary complex including quinolone, and thus could decrease the quinolone susceptibility of M. tuberculosis DNA gyrase. The M74I substitution abolishes the interaction between the residues at positions 74 in the dimer and makes the positioning of DNA in the DNA gyrase pocket possible. This is supported by the measurement of a similar IC₅₀ for the E. coli DNA gyrase and the altered M. tuberculosis gyrase A^M74I+A83SB. However, contrary to the results obtained in the supercoiling assay, the substitution at position 74 did not improve quinolone-induced cleavage. This could result from either the differences in binding conformation required for supercoiling inhibition and the induction of cleavage or from the fact that DNA gyrase does not need the segment of DNA to be well positioned in the DNA pocket to cleave it, although the enzyme does for better supercoiling activity.

FIG. 3.

Modeling of the 3-D structure of the 59-kDa N-terminal domain of GyrA of M. tuberculosis DNA gyrase on the basis of the corresponding domain of the E. coli DNA gyrase (19). (A) Ribbon representation of the 59-kDa N-terminal domain of GyrA of the E. coli DNA gyrase. (B) Close-up of the region outlined by the broken box, highlighting the positions of residues for the E. coli DNA gyrase (left and green) and for the M. tuberculosis DNA gyrase (right and gray). The active site (Tyr¹²²) is in blue, and the residues that were mutated in this study are in magenta. The numbering is according to the system of E. coli. The figure was generated using PyMOL (http://www.pymol.org).

Role of the GyrB QRDR.

We demonstrated that only the R447K substitution conferred increased quinolone susceptibility. Its association with A83S in GyrA, which is responsible for lower susceptibility to quinolones, was synergistic for nalidixic acid and, to a lesser extent, ofloxacin. The C-terminal domain of GyrB contains the QRDR (7) and is composed of a Toprim domain, which is also found in some primases, nucleases, and DNA repair proteins (1), and a “tail” domain (8). Amino acids 426 and 447 of GyrB are part of the “quinolone binding pocket,” along with DNA and the QRDR of GyrA (13). In E. coli, the positively charged K447 is hypothesized to interact with negatively charged D426, resulting in a neutral environment in the quinolone binding pocket, which would favor the binding of hydrophobic groups such as the nalidixic acid methyl group (13). Negatively charged D426 could also interact with the positively charged amino groups of the radicals located at R-7 of enoxacin, ofloxacin, and moxifloxacin. In M. tuberculosis, the environment of the quinolone binding pocket is positively charged, since D426 and R447 contribute two positive charges. This reduces the binding of nalidixic acid, due to its hydrophobic group, and of enoxacin, ofloxacin, and moxifloxacin, due to their positively charged R-7 radicals. Therefore, R447K could lead to the quinolone susceptibility we observed.

The GyrB QRDR may also be involved in quinolone trapping, since the mutated enzyme with R447K in GyrB cannot cleave DNA in the presence of quinolones. Recent studies have shown that point mutations in DNA gyrase subunit domains may modify enzymatic function (15). Further studies will be necessary to refine our understanding of the involvement of residue 447 in cleavage or religation.