Structural and Biochemical Analysis of the Pentapeptide Repeat Protein EfsQnr, a Potent DNA Gyrase Inhibitor

Abstract

The chromosomally encoded Qnr homolog protein from Enterococcus faecalis (EfsQnr), when expressed, confers to its host a decreased susceptibility to quinolones and consists mainly of tandem repeats, which is consistent with belonging to the pentapeptide repeat family of proteins (PRPs). EfsQnr was cloned with an N-terminal 6× His tag and purified to homogeneity. EfsQnr partially protected DNA gyrase from fluoroquinolone inhibition at concentrations as low as 20 nM. EfsQnr inhibited the ATP-dependent supercoiling activity of DNA gyrase with a 50% inhibitory concentration (IC₅₀) of 1.2 μM, while no significant inhibition of ATP-independent relaxation activity was observed. EfsQnr was cytotoxic when overexpressed in Escherichia coli, resulting in the clumping of cells and a loss of viability. The X-ray crystal structure of EfsQnr was determined to 1.6-Å resolution. EfsQnr exhibits the right-handed quadrilateral beta-helical fold typical of PRPs, with features more analogous to MfpA (mycobacterium fluoroquinolone resistance pentapeptide) than to the PRPs commonly found in cyanobacteria.

Quinolones are broad-spectrum synthetic antibacterials used extensively in the treatment of a variety of bacterial infections (14). They exert their powerful bactericidal activity by interacting with type II topoisomerases, namely, DNA gyrase and DNA topoisomerase IV. Type II topoisomerases cleave both strands of DNA during the catalytic cycle and allow the passage of a double-stranded segment of the same DNA molecule or of another molecule through the open gate before the original strands are religated. Quinolones stabilize the transient cleavage complex and block the religation; the subsequent hydrolysis of the enzyme-DNA bond results in the release of lethal double-stranded DNA breaks (9, 13, 17, 21). Gyrase is responsible for introducing and maintaining negative supercoils. Gyrase also can remove both positive and negative supercoils and catenate/decatenate closed-circular DNA molecules. Topoisomerase IV also removes both negative and positive supercoils and catalyzes decatenation more efficiently than gyrase; however, it lacks the ability to introduce supercoils into DNA. These two enzymes work together in the replication, transcription, and repair of DNA (6).

Bacteria have developed a variety of resistance mechanisms to evade the bactericidal effect of quinolones. Chromosomal mutations of the target gene resulting in amino acid substitution(s) in gyrase and/or topoisomerase IV have been the most clinically significant resistance mechanism (8, 14-16) before the recent emergence of horizontally transmissible plasmid-mediated quinolone resistance (PMQR) (19, 22, 24, 26, 37, 43, 44, and references therein), although other resistance mechanisms, such as decreased intracellular accumulation due to active efflux either alone or together with the decreased expression of outer membrane porins, have been described already (7, 23, 25, 27, 36). PMQR, first described in 1998 in an isolate of Klebsiella pneumoniae (28), was found to be due to a gene named qnr that encoded the 218-amino-acid Qnr protein belonging to the pentapeptide repeat protein (PRP) family. Two other PMQR mechanisms other than the expression of qnr, namely, an active efflux pump protein, QepA (33, 51), and an aminoglycoside N-acetyltransferase variant, AAC(6′)-Ib-cr, which catalyzes the acetylation of certain quinolones (29, 38, 50), also have been discovered.

In the last 5 years, several new qnr alleles and variants have been discovered worldwide from clinical isolates of Gram-negative pathogens, and the number of reported alleles/variants is close to 30 (4, 18). In addition, several pathogenic Gram-positive bacteria carry a chromosomal copy of qnr homologs, whose expression confers significantly reduced susceptibilities to quinolones in Escherichia coli (1, 39). The biochemical characterization of several qnr variants and homologs suggests that at least in part the molecular basis for qnr-mediated quinolone resistance is the binding of these proteins to gyrase and/or topoisomerase IV (45, 46), the inhibition of the topo-DNA interaction, and consequently the prevention of the quinolone-mediated cleavage complex stabilization. The differences in their abilities to inhibit gyrase/topoisomerase IV activities and to protect the gyrase from quinolone-mediated inhibition also suggests that there are additional mechanisms yet undiscovered. The three-dimensional structure of MfpA, a qnr homolog from Mycobacterium tuberculosis (MtMfpA), reveals a shape, size, and charge distribution reminiscent of a 30-bp long B-form DNA that would occupy the entire length of the G segment DNA binding saddle of type II topoisomerases (12). However, little is known about either the physiological role of plasmid-borne qnr variants and chromosomally encoded Gram-positive homologs or the physiological burden of carrying these gyrase inhibitors.

Enterococcus faecalis contains a chromosomal copy of a qnr homolog (EfsQnr) whose complementation/overexpression and heterologous expression (in both Gram-positive and Gram-negative bacteria) increased the MICs of various fluoroquinolones by 4- to 16-fold (1). In this paper, we describe the biochemical and kinetic characterization of gyrase inhibition and the three-dimensional structure of EfsQnr, and we report on the cytotoxic effects of EfsQnr overexpression in E. coli.

MATERIALS AND METHODS

Materials.

All chemicals and reagents were purchased from Sigma-Aldrich Chemical Co. Enzymes used in molecular biology were supplied by New England Biolabs. Plasmid pET-28a, E. coli strains Nova Blue and BL21(DE3), were obtained from Novagen. DNA gyrase, topoisomerase IV, and the assay kits were from Inspiralis.

Cloning, overexpression, and purification of EfsQnr.

The open reading frame of EfsQnr was amplified by standard PCR techniques using E. faecalis V583 (ATCC 700802) chromosomal DNA as the template. Oligonucleotides QEfF (5′-ATCCCGCTCATATGAAAATAACTTATCCCTTGCCA-3′) and QEfR (5′-ATCCCGCTCTCGAGTTAGGTAATCACCAAACCAAGT-3′), containing the underlined NdeI and XhoI restriction sites, respectively, were used. The PCR fragment was cloned into pET-28a(+), and the recombinant EfsQnr bearing a thrombin-cleavable N-terminal His₆ tag was expressed in E. coli strain BL21(DE3). For shake flask growth, 1 liter of Luria broth medium supplemented with kanamycin (35 μg/ml) was inoculated with 10 ml of an overnight culture and incubated at 37°C. The culture was grown to mid-log phase (A₆₀₀ ∼ 0.8), cooled to 20°C, induced with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and further incubated overnight at 20°C.

All purification procedures were carried out at 4°C. The cells were collected by centrifugation at 1,200 × g and resuspended in buffer A (50 mM Tris-HCl [pH 7.8], 300 mM NaCl) containing protease inhibitors, lysozyme (5 μg/ml), and DNase I (0.1 μg/ml), and the mixture was stirred for 20 min. The cells then were lysed by sonication, and cell debris was removed by centrifugation at 10,000 × g for 30 min. The supernatant was loaded onto a nickel-nitrilotriacetic acid (Ni-NTA) column preequilibrated with buffer A and washed with 10 column volumes of the same buffer. The bound proteins were eluted with a linear 0 to 0.3 M imidazole gradient with fractions pooled and concentrated to >20 mg ml⁻¹ by ultrafiltration. The His₆ tag then was cleaved using thrombin (2 U mg⁻¹ of protein), and the solution was dialyzed overnight against buffer A containing 2 mM CaCl₂ and loaded onto a Superdex S-75 column preequilibrated with buffer A. Pure fractions as determined by SDS-PAGE were pooled and concentrated to >20 mg ml⁻¹ by ultrafiltration.

Protein estimation.

Protein concentrations were estimated by the Bio-Rad protein assay method using bovine serum albumin as a standard. Analytical gel filtration was performed using a Superose 12 10/30 fast performance liquid chromatography (FPLC) column (Pharmacia) equilibrated with 50 mM triethanolamine (TEA), pH 7.8, containing 200 mM ammonium sulfate and 5% glycerol at a flow of 0.5 ml/min. Molecular mass standards (Bio-Rad) contained thyroglobulin (670 kDa), bovine gamma globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and vitamin B (1.35 kDa). Dynamic light scattering (DLS) was measured with a DynaPro MS/X dynamic light-scattering instrument (Protein Solutions) with EfsQnr at 10 mg ml⁻¹ in 10 mM Tris-HCl (pH 7.8), 20 mM NaCl, 1 mM dithiothreitol (DTT).

Gyrase and topoisomerase assays.

DNA supercoiling, relaxation, and decatenation assays were performed using E. coli gyrase and topoisomerase IV assay kits (Inspiralis) according to the manufacturer's instructions. For gyrase supercoiling assays, the reaction mixture containing 3 U (a unit is defined as the amount of gyrase required to convert 0.5 μg of relaxed pBR322 into the completely supercoiled form at 37°C in 30 min) of gyrase and 0.4 μg relaxed DNA in a volume of 30 μl in gyrase assay buffer (35 mM Tris-HCl [pH 7.5], 24 mM KCl, 4 mM MgCl₂, 2 mM DTT, 1.8 mM spermidine, 1 mM ATP, 6.5% glycerol, and 0.1 mg ml⁻¹ bovine serum albumin [BSA]) was incubated at 37°C for 30 min, and EfsQnr and/or ciprofloxacin was included where appropriate. Reactions were terminated by the addition of 30 μl chloroform-isoamyl alcohol (24:1). The resulting topoisomers were separated by agarose gel electrophoresis, stained with ethidium bromide, and visualized under UV light. Inhibition and/or protection was calculated from band intensities using Syngene Bioimage analysis software. Gyrase relaxation assays were carried out using 30 U of gyrase as described above, except that supercoiled pBR322 was used as the substrate and spermidine and ATP were omitted from the assay buffer. For decatenation assays, 2 U of E. coli topoisomerase IV was incubated for 30 min at 37°C with 0.2 μg kDNA in assay buffer (40 mM HEPES, pH 7.6, 100 mM potassium glutamate, 10 mM magnesium acetate, 10 mM DTT, 1 mM ATP, and 50 μg/ml BSA) in a total volume of 30 μl.

SEM.

For scanning electron microscopy (SEM) studies, the washed cell samples were fixed using 2.5% glutaraldehyde in 100 mM sodium cacodylate buffer, pH 7.4, containing 200 mM sucrose and 5 mM MgCl₂. Samples then were dehydrated through a graded series of ethanol washes, and critical point drying was carried out using liquid carbon dioxide in a Tousimis Samdri 795 critical point drier. Coating was performed with gold-palladium in a Denton vacuum desk-2 sputter coater, and the samples were examined using a JEOL JSM6400 scanning electron microscope at an accelerating voltage of 10 kV.

Crystallization.

The crystallization and structure determination of EfsQnr are described elsewhere (48). Briefly, the structure of EfsQnr was determined by single anomalous dispersion from a single samarium acetate-soaked crystal. The structure was determined to a resolution of 1.6 Å and refined to an R_factor and R_free of 18.7 and 22.1, respectively (48). There are two subunits per asymmetric unit, which combine to form the molecular dimer (see below). Electron density was observed for all residues except those resulting from the cleavage of the His₆-thrombin-cleavable tag (His⁻³, Ser⁻², Gly⁻¹) and residues 9 to 12 of the A subunit and residues 1 to 3 of the B subunit, both part of a flexible N terminus.

RESULTS

Purification and properties of EfsQnr.

PCR amplification of the EfsQnr gene yielded a single fragment of the expected length. The cloning and overexpression of the PCR product resulted in expressed protein product with an apparent molecular mass, by SDS-PAGE, in agreement with the mass deduced from the nucleotide sequence (∼24,250 Da). The DNA sequencing of the cloned fragment confirmed the absence of any mutations introduced during PCR amplification. The two-step purification procedure yielded homogeneous protein as determined by SDS-PAGE. Analysis by DLS and analytical gel filtration suggests EfsQnr is a molecular dimer in solution. DLS of EfsQnr yielded a monodisperse peak with a radius of gyration of 29 Å, which is consistent with a molecular mass of 40 kDa for a spherical particle. EfsQnr exhibited a single peak in analytical gel filtration and coeluted with ovalbumin (44 kDa) when injected with a set of molecular mass standards. The evidence of a molecular dimer in solution is consistent with the crystal data, which illustrates a C-terminal dimer interface similar to that observed with MtMfpA (see below).

Inhibition of DNA gyrase and protection against quinolone inhibition by EfsQnr.

Purified EfsQnr inhibited the ATP-dependent DNA supercoiling activity of E. coli gyrase with a calculated 50% inhibitory concentration (IC₅₀) of 1.2 μM (Fig. 1A). However, the ATP-independent relaxation activity of the gyrase was poorly inhibited (Fig. 1B) even at EfsQnr concentrations as high as 100 μM. EfsQnr partially protected gyrase against ciprofloxacin inhibition, and partial protection was observed even at the lowest tested concentration of 20 nM (data not shown). The determined IC₅₀ for gyrase inhibition by ciprofloxacin was 0.25 μM, while in the presence of 0.2 μM EfsQnr the IC₅₀ increased to 1.4 μM (Fig. 2). However, EfsQnr did not exhibit any inhibitory effect on decatenation and relaxation activities catalyzed by E. coli topoisomerase IV (data not shown). A structure-based deletion mutant (see below), Δ17EfsQnr_, lacking the first 17 residues from the N-terminal end, was expressed and purified. Δ17EfsQnr inhibited gyrase supercoiling activity poorly (see Fig. S1 in the supplemental material), and its ability to protect gyrase from quinolone inhibition was reduced drastically (data not shown). The calculated IC₅₀ of 23 ± 5 μM was about 20-fold higher than that for full-length EfsQnr. Δ17EfsQnr, like full-length EfsQnr, inhibited neither ATP-independent relaxation activity of gyrase nor decatenation and relaxation activities of topoisomerase IV (data not shown).

FIG. 1.

Inhibition of DNA gyrase by EfsQnr. Inhibition of supercoiling (A) and relaxation activities (B) of DNA gyrase by EfsQnr. (A) Lane 1, relaxed plasmid pBR322 alone; lane 2, relaxed pBR322 plus 3 U of gyrase; lanes 3 to 8, 3 U of gyrase and 0.25, 0.5, 1, 2, 5, and 10 μM EfsQnr, respectively. (B) Lane 1, supercoiled plasmid pBR322 alone; lane 2, supercoiled pBR322 plus 50 μM EfsQnr; lane 3, supercoiled pBR322 plus 25 U of gyrase; lanes 4 to 8, 25 U of gyrase and 5, 10, 25, 50, and 100 μM EfsQnr, respectively. nc, l, and sc represent nicked circular, linear, and supercoiled forms, respectively.

FIG. 2.

Inhibition of supercoiling activity by ciprofloxacin and partial protection by EfsQnr. (A) Lane 1, relaxed plasmid pBR322 alone; lane 2, relaxed pBR322 plus 3 U of gyrase; lanes 3 to 8, 3 U of gyrase and 0.1, 0.25, 0.5, 1, 2, and 5 10 μM ciprofloxacin in the absence EfsQnr (top) and in the presence of 0.2 μM EfsQnr (bottom), respectively. (B) Graphical representation of the inhibition data. Symbols are experimentally determined values, while the smooth lines are the fit of the data.

Effect of EfsQnr expression on E. coli.

Signs of cytotoxicity were observed in E. coli cultures expressing EfsQnr. SEM studies were performed as described in Materials and Methods. Samples were taken at different time intervals after induction. Cells expressing EfsQnr and Δ17EfsQnr exhibited a distinct change in their shape, although the size remained the same compared to that of the control (Fig. 3A). Upon induction, cells were able to divide but did not separate. Daughter cells were connected through an external appendage-like structure, forming clusters (Fig. 3B). The number of cells in clusters increased as a function of time, and after 8 h of induction several hundred cells were seen in these clusters (Fig. 3C). Many of the cells carried various numbers of node-like appendages on their surfaces (Fig. 3B). When an induced culture was plated on LB agar plates, more than 98% of the cells (based on A₆₀₀) had lost their viability (colony forming ability) compared to the control (Fig. 3D). The restriction digestion of the plasmids isolated from the plated colonies revealed that half of the colonies that were formed did not carry the EfsQnr gene insert (data not shown).

FIG. 3.

Scanning electron microscopy of E. coli BL21(DE3) cells expressing EfsQnr. (A) Control cells; (B) 2 h after induction of EfsQnr expression; (C) 8 h after induction. (D) Effect of EfsQnr induction on the viability of E. coli. Cultures were diluted identically and plated onto LB plates containing 30 μg/ml kanamycin. Bottom, control; top, cells expressing EfsQnr.

Crystal structure of EfsQnr.

The pentapeptide repeats of EfsQnr folds as a right-handed quadrilateral β-helix as observed for other members of the PRP family. Each repeat occupies one face of a quadrilateral, with each coil consisting of 20 residues and with each revolution of the β-helix traversing approximately 4.8 Å. The side chains of the i and i⁻² residues (see Fig. 4 for nomenclature) almost invariably point toward the center of the β-helix making up its hydrophobic core. Similarly, the side chains of the i⁻¹, i⁺¹, and i⁺² residues project out into solvent and make up the surface of the molecule. The pentapeptide sequence also can be envisioned as β-strand elements connected by either type II or type IV turns. In the case of a type II turn, the carbonyl of the i residue is hydrogen bonded to the i⁻² residue of the following repeat, and only i⁻¹ is in a full β-strand interaction (β-bridge; spheres in Fig. 5A). Alternatively, in the type IV turns of PRPs the plane of the peptide between the i and the i⁻¹ residues rotates 90°, such that i and i⁻¹ also make full intercoil β-strand interactions (strands in Fig. 5A). Due to the alternative hydrogen bonding, coil faces with pentapeptides preceding type IV turns are approximately 0.5 to 0.75 Å longer than those that utilize β-bridges and precede type II turns.

FIG. 4.

Primary sequence of EfsQnr and MtMfpA. Pentapeptide repeats are compiled into four columns (gray-boxed residues) indicating their location within the four faces of the quadrilateral β-helix. Pentapeptide residue types (i.e., i⁺¹, i⁺², etc.) are shown between the two sequences. Helices are displayed in salmon-colored boxes. Residues that support the largest deviations from the standard β-helix are shown in blue.

FIG. 5.

Monomer structure of EfsQnr and MtMfpA. (A) Ribbon diagram of EfsQnr and MtMfpA. β-Helical structural elements are colored by face type as aligned in Fig. 4. α-Helices are in salmon, and the N-terminal extension of EfsQnr is black. Coils have repeats that are colored in the order green, cyan, yellow, and red. Residues within repeats that exhibit full intercoil hydrogen bonding are shown as strands and are indicative of the following turn being type IV, while β-bridges are illustrated as spheres and are indicative of the following turn being type II. (B) Stereoview of the superposition of EfsQnr (black trace) and MtMfpA (orange trace).

β-Helix capping.

The numbering of the EfsQnr coils is based on that of the functionally analogous protein MtMfpA with an orientation anchored by a highly similar dimer interface at the C terminus. EfsQnr exhibits six uninterrupted regular coils (C1 to C6) with partial and or modified coils at the N and C terminus (Fig. 5A, C0, C7 to C9). PRPs and other β-helices typically are capped on their ends to prevent aggregation with the β-strand elements of other proteins. The N-terminal end of the EfsQnr β-helix is capped by residues that are atypical of PRP sequences (such as His32 and Gln42 at the i⁻² residue) in addition to a peptide (residues 19 to 26) that transverses the diagonal of the β-helix (face 1/face 2 turn to face 3/face 4 turn). In the process, Cys21 and Leu23 of the peptide interact with the hydrophobic environment of the interior of the terminal coil (see Fig. S2 in the supplemental material). The C-terminal end of the β-helix is covered by the dimerization domain, which includes strand (coil 9/face 1), a helix (α2), and a strand exchange peptide.

PRP sequence deviations.

Most of the protein core, comprised of the i and i⁻² residues, follows the typical PRP sequence, with a small polar or hydrophobic residue at the i⁻² residue and a leucine or phenylalanine at the i residue. The analysis of the deviations from the consensus sequence among a body of PRP structures provides predictive power for modeling PRP sequences that lack a structure. There are two regions of interesting deviation from the consensus sequence, and both involve tryptophan residues at the i position. Coil 3, face 3 has a tryptophan at the i position with the indole ring buried in the hydrophobic interior. The indole nitrogen does not form a specific hydrogen bond with a protein residue but instead interacts with the pi-electron cloud of the side chain of Phe67 (see Fig. S3A in the supplemental material). Phe67 is in the i⁻² position of face 4, coil 2, and its presence must be accommodated by a smaller residue at the i position of the same face (Cys68). The side chain of Phe64 (i residue, coil 2/face 3) takes a nontypical PRP conformation to prevent collisions with Phe67 and Trp84. This alternate conformation, which takes the side chain away from the central cavity and toward the face 2/face 3 turn, is facilitated by the glycine at the i⁻² position of coil 2/face 3. The bulkiness of Trp83 also is facilitated by a slight separation of coil 2 from coil 3 at the face 3/face 4 turn. For example, the coil 2/coil 3 intercoil distance at the i⁺² residue expands from 4.8 to 6.3 Å, and while PRP interiors typically are devoid of water, in this case a water molecule is captured between coil 2 and 3, forming a hydrogen bond between Phe64^O and Phe67^N. Similar waters are captured in the turns between face 3/face 4 on coil 1 (Val44^O-H₂O-Leu47^N) and face 2/face 3 of coil 1 (Leu39^O-H₂O-Gln42^N) (see Fig. S3A in the supplemental material). This demonstrates that internally bound waters can be utilized to stabilize nontypical PRP sequences.

The second tryptophan at an i position is Trp149, and its presence helps support the largest deviation in the β-helical structure on face 3/face 4 in coils 6 to 8 (see Fig. S3B). Instead of projecting into the center of the β-helix, the side chain of Trp149 projects upwards to make an intercoil hydrogen bond between the indole nitrogen (2.8 Å) and the carbonyl of Ser168. This increases the intercoil separation to greater than 9 Å at the i⁺² position of face 4, coil 6/coil 7. The coil separation is further supported by uncharacteristic PRP residues in adjacent faces, such as Trp164 at an i position and Leu152, Leu172, and Asn177 at i⁻² positions. Asn177 may be especially important, as its side chain makes two intercoil bifurcated main-chain hydrogen bonds (Asn177^OD1-Ala156^N/Cys157^N and Asn177^ND2-Asp173^O/Leu154^O), maintaining intercoil connectivity but with a larger intercoil distance, while other main-chain atoms exposed due to intercoil separation are solvated by a number of waters.

Interestingly, both tryptophans are at sites of transition from a face with type IV turns to a face with type II turns. Typically, the type of turn on a particular face is conserved as one transitions along the β-helix, as this optimizes intercoil hydrogen bonding. The use of unconventional residues at internal positions therefore may be used to break the intercoil hydrogen bonding, provide a region of flexibility, and transition to an alternative intercoil hydrogen-bonding scheme.

N-terminal extension.

Residues 1 to 18 are not part of the β-helical fold but extend as a random coil interacting with a groove along the β-helix created by the outward-facing side chains of the i⁺² residue of face 4 and the i⁻¹ residue of face 1 (Fig. 5A; also see Fig. S2 in the supplemental material). The N termini of the two subunits make dissimilar contacts with this groove, and it is not clear if either conformation is biologically relevant; however, the N termini of subunit B has more specific contacts, while the N terminus of subunit A forms an intramolecular contact with a lysyl-piperidine modification (Lys2) that is not visualized in subunit B and would not be physiological. A number of hydrophobic residues from the N terminus (Leu11, Leu15, Pro6, Pro9) are buried within the groove, while all of the polar contacts are β-helical side chains to main-chain atoms of the N-terminal extension. These include Arg91^NE/NH1-Pro8^O, Asn71^OD1-Asn10^N, and Asp111^OD1-Leu7^N. This is the first instance of a PRP structure that includes extensive PRP residues N terminal to the β-helix that are not utilized in the capping of the N terminus.

Dimer interface.

There are two subunits per asymmetric unit with the molecular dimer created by noncrystallographic symmetry. The subunits are adjoined at their C termini in such a fashion that the two β-helical axes are nearly colinear, creating a cylindrically shaped dimer with an axis of approximately 110 Å and a diameter that varies between 30 Å (N-terminal coils) and 15 to 20 Å (C-terminal dimer interface) (Fig. 6). The dimer interface is composed mainly of residues from the N-terminal α helix (α2) and a crossover exchange where residues 209 to 211 form a β-strand on face 1 of the symmetry-related β-helix (parallel β-strand). Outside of the polar interactions formed by the C-terminal β-strand the majority of the dimer interface is composed of buried hydrophobic side chains (L189, L192, V194, A199, I200, L202, A203, L206, L208). A total of 930 Ų angstroms is buried per subunit in the dimer interface. The two subunits, outside the mobile N termini, are highly similar, with a root mean square deviation (RMSD) of 0.36 Å over 193 common Cα atoms. The diminutive nature of the dimer interface in addition to its apolar nature appears to permit flexibility in the relative orientation of the two subunits. In a global superposition of the subunits the C-terminal residues are displaced relative to each other, yielding an RMSD of 0.69 (residues 194 to 211). This results in an imperfect 2-fold, which starting from the C-terminal dimer interface gets amplified as it travels down the β-helix, with Cα translations in the N-terminal coil of up to 3 Å from perfect symmetry. The asymmetry is presumably held in place through asymmetrical crystal contacts to the individual subunits.

FIG. 6.

Molecular dimer of EfsQnr. Ribbon diagram of the EfsQnr and MtMfpA dimer orientated based on the superposition of MtMfpA on subunit B of EfsQnr.

DISCUSSION

Comparison to MtMfpA.

The pentapeptide protein MtMfpA is intriguingly similar in many ways to EfsQnr. Both MtMfpA and EfsQnr are chromosomally encoded and have an unknown biological function, but they are known to provide the host with a decreased sensitivity to fluoroquinolones. Both are approximately 200 amino acids in length and almost entirely composed of tandemly repeated 5-amino acid sequences, resulting in the main body of their structures being a quadrilateral β-helix. Both are dimers formed through a relatively small C-terminal domain consisting of a single α-helix and a domain-swapped β-strand. The dimeric structures are highly nonspherical, with a nearly colinear β-helical axis of approximately 110 Å. Despite these similarities, MtMfpA and EfsQnr exhibit a pairwise sequence identity of only 19.5% (32.3% similarity; PAM250 matrix), which is surprisingly low when one considers the intrinsic sequence restrictions imposed by the PRP consensus sequence. EfsQnr is different in some respects. EfsQnr has an altered N-terminal structure, with an additional β-helical rung on face 4, an N-terminal capping peptide (residues 19 to 26), and a mobile N-terminal extension (residues 1 to 25), while MtMfpA has no structures N terminal to the β-helix. MtMfpA has a change in the helical axis between coils 4 and 6 that disrupts the intercoil hydrogen bonding between coil 4/coil 5 on face 4 and coil 5/coil 6 on face 1. EfsQnr only has a minor β-helical disruption between coil 2/coil 3 on face 2, due to a buried tryptophan, with relatively little change in the helical axis for the lower coils. As such, a superposition of MtMfpA and EfsQnr based on the lower three coils results in a large deviation between the two structures for coils 4 to 7 on faces 4 and 1. Interestingly, the disruption of the intercoil hydrogen bonding in coils 6 to 8 of EfsQnr on faces 4 and 1 results in a correction in the registration between the two structures and they resume a structural overlap (Fig. 5B). The effect is a similarity in the orientation of the lower three coils relative to the C-terminal residues. In addition, there is very little correlation between MtMfpA and EfsQnr as to the location of type II versus type IV turns. For example, in EfsQnr the face 4 to face 1 transition is entirely composed of type IV turns, while in MtMfpA it is entirely composed of type II turns. The only correlation is a larger proportion of the lower coils contain type IV turns while the upper coils tend to contain type II turns.

MtMfpA was proposed to interact with the saddle region of the GyrA molecular dimer, which has a positive electrostatic character (12, 31). Both MtMfpA and EfsQnr exhibit molecular surfaces dominated by electronegative patches that are consistent with interacting with a DNA-binding region of gyrase (see Fig. S4 in the supplemental material). EfsQnr exhibits a more-electronegative surface, as suggested by its larger net negative charge (MtMfpA, −50 [+42 to −8]; EfsQnr, −54 [+32 to −22]). While general electrostatic surface matching between EfsQnr (or MfpA) with gyrase will play an important role, there presumably are more specific interactions as the structure of PRPs, which do not inhibit DNA gyrase, also are right handed β-helices with overall negative surface character.

MtMfpA and EfsQnr differ in several aspects from PRPs that do not inhibit DNA gyrase. MtMfpA and EfsQnr and related PRP DNA gyrase inhibitors have a more-diverse set of sequences within their repeats than those of non-DNA gyrase inhibitors (2, 3, 49). For example, MtMfpA and EfsQnr are more likely to have a cysteine, threonine, or valine at the i⁻² position than the average PRP that is highly likely to have an alanine at the i⁻² position. Similarly, the i⁻¹ and i⁺¹ sequences of typical PRPs have a preference for Asp/Asn and Thr/Ser/Arg, respectively, as an Asp/Asn side chain of an i⁻¹ residue can hydrogen bond to the amide of the i⁺¹ residue when the i⁺¹ residue participates in a type II turn and Thr/Ser/Arg at the i⁺¹ residue then can form a hydrogen bond to the i⁻¹ side chain. In contrast, MtMfpA and EfsQnr have fewer of the i⁻¹-i⁺¹ interactions and therefore have a more diverse set of sequences at these surface positions. The structures of PRPs that are not DNA gyrase inhibitors contain predominantly type II turns and are very uniform, with no kinks in the β-helix. Presumably, with fewer transitions between adjacent type IV and type II coils, there is less need for nonstandard amino acids at the i and i⁻² positions to break the intercoil hydrogen bonding. Perhaps most importantly, while MtMfpA and EfsQnr are dimers, the non-DNA gyrase inhibitors all have been monomeric. It is not clear how the structural differences between the simpler and more complex PRPs are translated into the ability to inhibit DNA gyrase. Clearly, the dimerization of MtMfpA and EfsQnr suggests a more-potent interaction across a molecular target that is also a dimer, and therefore dimerization may be an important feature. It is beguiling how the quadrilateral β-helix of MtMfpA and EfsQnr can invoke a similar inhibition with a complete lack of sequence conservation in the outward-facing residues. Further research involving the mapping of the MtMfpA-EfsQnr DNA gyrase interface is required to elucidate the important molecular features of this interaction.

Almost all of the qnr variants and homologs, when expressed, are capable of conferring quinolone resistance in vivo, albeit at various levels. However, they seem to differ in their abilities to inhibit gyrase. MtMfpA inhibits both supercoiling and relaxation activities of gyrase but does not protect gyrase from quinolone inhibition (12, 30). QnrA₁ binds and protects both gyrase and topoisomerase IV from quinolone inhibition but does not inhibit the enzyme activities (45, 46). QnrB₁ protects gyrase from quinolone inhibition at picomolar (as little as 5 pM) concentrations, while micromolar amounts are required for gyrase inhibition (20). In contrast, EfsQnr inhibits supercoiling activity of gyrase at low-micromolar concentrations as well as partially protects gyrase from quinolone inhibition at nanomolar concentrations. Partial protection against ciprofloxacin inhibition was observed at a concentration as low as 20 nM. The highest levels of protection were observed at EfsQnr concentrations between 0.1 and 0.2 μM, above which EfsQnr inhibited gyrase significantly. The calculated IC₅₀ of gyrase inhibition by ciprofloxacin increased by 6-fold in the presence of 0.2 μM EfsQnr (the concentration at which inhibition due to EfsQnr is less than 10%). This increase roughly coincides with 4- to 8-fold increases in MICs for ciprofloxacin upon the complementation and/or heterologous expression of EfsQnr in different Gram-negative and Gram-positive organisms (1). MtMfpA inhibited both supercoiling and relaxation activities of the gyrase with identical IC₅₀s, while EfsQnr exhibited a great difference in its ability to inhibit ATP-dependent supercoiling and ATP-independent relaxation activities of the gyrase. No significant inhibition of ATP-independent relaxation was observed at concentrations as high as 100 μM. Interestingly, EfsQnr failed to inhibit ATP-dependent relaxation and decatenation catalyzed by E. coli topoisomerase IV. In a recent finding it was reported that MtMfpA inhibited both E. coli and M. tuberculosis gyrases but failed to protect these gyrases from quinolone inhibition, while QnrB4 reversed the quinolone-mediated inhibition of E. coli gyrase but not of M. tuberculosis gyrase, indicating that qnr-mediated inhibition/protection are species and protein specific (30).

Plasmid-borne qnr variants appear to have been derived from chromosomal genes from environmental species (5, 34, 35, 40, 42, 47). Neither the physiological role of chromosomally encoded qnr homologs and plasmid-borne variants/alleles nor the physiological burden of carrying potential gyrase inhibitors is known. EfsQnr has a unique structural feature in that its first 18 residues form a random coil that extends along the surface groove, making several contacts. To ascertain the role of this structural element, we expressed an N-terminal deletion mutant, Δ17EfsQnr. Purified Δ17EfsQnr was a much weaker inhibitor of gyrase, exhibiting a 20-fold higher IC₅₀ for gyrase and no inhibitory effect on topoisomerase IV. However, visible signs of toxic effects, such as longer doubling times and lower cell densities, were observed in E. coli strains expressing Δ17EfsQnr. Microscopic examination revealed that the E. coli strains expressing both EfsQnr and Δ17EfsQnr exhibited similar morphological changes; however, under identical expression conditions Δ17EfsQnr expression was at least 10-fold lower than that of full-length EfsQnr. Morphological changes associated with fluoroquinolone and other DNA gyrase inhibitors have been studied in E. coli and other bacteria (39-42). Fluoroquinolones and other gyrase inhibitors cause the filamentation of E. coli and interfere with nucleoid segregation (10, 11, 32, 41). Morphological changes due to the presence of sub-MIC levels of fluoroquinolones are reversible upon the removal of the drug. However, the morphological changes due to EfsQnr (and Δ17EfsQnr) are distinct; no filamentation of the cells was observed. These data indicate that EfsQnr, apart from inhibiting gyrase, inhibits additional yet-unknown cellular processes, most probably the final steps in cell division leading to the separation of daughter cells. Interestingly, the overexpression of EfsQnr does not appear to be bactericidal for actively growing E. coli. However, after a few cell cycles upon induction (overexpression), cells lose their ability to form colonies. It remains to be seen at which point during this expressing stage they reach the critical point of nonviability. It is reasonable to speculate that the bactericidal effects of EfsQnr were not primarily due to the inhibition of DNA gyrase, as this would have inhibited the DNA replication and nucleoid segregation, leading to filamentation.

To the best of our knowledge, we have for the first time demonstrated the bactericidal effect of a qnr homolog. It is possible that other qnr homologs and plasmid-borne qnr variants exert similar cytotoxic effects in E. coli and other enteric bacteria. Given the differences in their abilities (or the lack thereof) to inhibit gyrase, protect gyrase from quinolone inhibition, and cause various levels of quinolone resistance, it is possible that some qnr variants are more potently bactericidal than others, and this effect may be species specific. The differences in MICs of quinolones against qnr-expressing strains are believed to be due to several reasons (26), such as differences in copy numbers of the plasmids that carry them, variations in transcriptional levels, the promoter strength of the gene carried by different plasmids, and the presence of additional resistance mechanisms. The quinolone-induced expression levels of qnr proteins in quinolone-resistant strains and E. coli transconjugants or the natural expression of chromosomally encoded qnr homologs may not reach the toxic levels observed under IPTG-induced overexpression. Further work is required to test whether qnr-induced cytotoxicity could be exploited to overcome this mechanism of fluoroquinolone resistance mechanism that bacteria have developed.