Identification of the Enzyme Responsible for N-Acetylation of Norfloxacin by Microbacterium sp. Strain 4N2-2

Abstract

Microbacterium sp. 4N2-2, isolated from a wastewater treatment plant, converts the antibacterial fluoroquinolone norfloxacin to N-acetylnorfloxacin and three other metabolites. Because N-acetylation results in loss of antibacterial activity, identification of the enzyme responsible is important for understanding fluoroquinolone resistance. The enzyme was identified as glutamine synthetase (GS); N-acetylnorfloxacin was produced only under conditions associated with GS expression. The GS gene (glnA) was cloned, and the protein (53 kDa) was heterologously expressed and isolated. Optimal conditions and biochemical properties (K_m and V_max) of purified GS were characterized; the purified enzyme was inhibited by Mn²⁺, Mg²⁺, ATP, and ADP. The contribution of GS to norfloxacin resistance was shown by using a norfloxacin-sensitive Escherichia coli strain carrying glnA derived from Microbacterium sp. 4N2-2. The GS of Microbacterium sp. 4N2-2 was shown to act as an N-acetyltransferase for norfloxacin, which produced low-level norfloxacin resistance. Structural and docking analysis identified potential binding sites for norfloxacin at the ADP binding site and for acetyl coenzyme A (acetyl-CoA) at a cleft in GS. The results suggest that environmental bacteria whose enzymes modify fluoroquinolones may be able to survive in the presence of low fluoroquinolone concentrations.

INTRODUCTION

Fluoroquinolones are widely used as human and veterinary antimicrobial agents (1). Because their persistence in the environment (2) may act as a selective pressure for naïve strains to acquire resistance (3), an understanding of the fate of these drugs should help to prevent drug resistance. The principal resistance mechanisms for fluoroquinolones include the following: (i) mutation of genes for the drug targets (GyrA, GyrB, and ParC) (4–6), (ii) mimicking the drug targets (QnrA, QnrB, and QnrS) (7–9), (iii) reduction of drug accumulation in the cells (AcrAB-TolC and QepA) (4, 10, 11), and (iv) enzymatic modification of the drug [N-acetylation by AAC(6′)-Ib-cr] (12–14).

In some strains of Escherichia coli, a mutated aminoglycoside acetyltransferase, AAC(6′)-Ib-cr, acetylates fluoroquinolones at the N-terminal of the piperazine ring (12–14) and enhances bacterial resistance to these drugs (14). Production of N-acetylnorfloxacin and N-nitrosonorfloxacin by environmental Mycobacterium sp. strains has also been found, but the enzymes responsible for modifications are unknown (15) and the aac(6′)-Ib-cr variant gene has not been detected in these strains (12).

A norfloxacin-modifying bacterium, Microbacterium sp. strain 4N2-2, was isolated from wastewater (16). Some strains of this genus have been isolated from human clinical specimens, and the majority of them have shown fluoroquinolone resistance (17). Microbacterium sp. 4N2-2 transforms norfloxacin into four different metabolites, including N-acetylnorfloxacin, and cell extracts of this strain catalyze the N-acetylation of norfloxacin (16). N-Acetylation is enhanced by Casamino Acids and inhibited by ammonium and nitrate (16), suggesting that the enzyme responsible for norfloxacin N-acetylation might also be involved in nitrogen metabolism.

In the present study, an enzyme with N-acetyltransferase activity was isolated from Microbacterium sp. 4N2-2 and identified. The coding gene was cloned and heterologously expressed in E. coli for biochemical characterization of the enzyme and its contribution to norfloxacin resistance. Binding models of the enzyme with norfloxacin and a cofactor were proposed by in silico structure modeling and ligand binding (docking) analysis. The results provide insight into the potential for enzymes of environmental bacteria to contribute to drug resistance.

MATERIALS AND METHODS

Strains, chemicals, and media.

Microbacterium sp. strain 4N2-2 was isolated from a wastewater treatment plant in Little Rock, AR (16). E. coli BL21(DE3) pLysS, for heterologous overexpression of the target protein, was purchased from Novagen (EMD Millipore, Billerica, MA). Both strains were stored in 20% glycerol at −80°C.

Norfloxacin (Sigma-Aldrich, St. Louis, MO) was prepared as a 10 mg ml⁻¹ stock solution in 40 mM KOH for cultures and enzyme assays.

OM medium (16) was used as the basal medium for cultures of Microbacterium sp. 4N2-2. For extraction of proteins, cultures were grown in OM medium with 2.0 g liter⁻¹ Casamino Acids instead of ammonium and nitrate (16). To test the effects of nitrogen sources on N-acetylation, 1.0 g liter⁻¹ of glutamate, glutamine, or ammonium was substituted. Cultures of Microbacterium sp. 4N2-2 were incubated at 30°C with shaking at 200 rpm for 9 days for protein extraction or for 10 to 17 days for N-acetylation analysis.

LB broth with 10 g liter⁻¹ NaCl (BD Biosciences, Franklin Lakes, NJ) was used for cultures of E. coli, and LB agar (20 g liter⁻¹ agar) was used for the norfloxacin disk assay. Cultures of E. coli were incubated aerobically at 37°C with shaking at 200 rpm for 18 h.

High-performance liquid chromatography.

After cultures were centrifuged (13,000 × g) and filtered (0.45 μm pore size), 10 μl of the spent medium was directly injected for high-performance liquid chromatography (HPLC) analysis of norfloxacin N-acetylation at 280 nm as previously described (16).

N-Acetylation by cell extracts.

After incubation in OM medium with Casamino Acids for 10 to 17 days, cells of Microbacterium sp. 4N2-2 were washed twice and suspended in 50 mM HEPES buffer (pH 7.5) containing 10% glycerol. A 4-ml volume of the cell suspension was disrupted by five passages through a French pressure cell at 137,000 kPa. Cell debris was removed by centrifugation (13,000 × g) and filtration (0.22 μm pore size). Various pH values, temperatures, and amounts of enzyme, substrate, and cofactors were used to evaluate the optimum conditions for the reaction. Cell extracts were incubated with 0.06 mM norfloxacin–0.2 mM acetyl coenzyme A (acetyl-CoA) (14)–5.0 mM CaCl₂–50 mM Tris-HCl buffer (pH 8.0) at 45°C. Production of N-acetylnorfloxacin was monitored by direct HPLC analysis of incubated reaction mixtures and controls.

Purification of the enzyme responsible for N-acetylation.

Ammonium sulfate was added to cell extracts to reach a final concentration of 1.0 M; the mixture was centrifuged and filtered (0.45 μm pore size). The filtrate was applied to a HiPrep phenyl FF 16/10 column (GE Healthcare, Piscataway, NJ). The proteins were eluted using 60 ml of 1 M (NH₄)₂SO₄, 60 ml of 0.5 M (NH₄)₂SO₄, 60 ml of 0.2 M (NH₄)₂SO₄, and 60 ml of water. The active fractions [eluted by the use of 0.2 M (NH₄)₂SO₄], as shown by HPLC, were collected, concentrated, and diluted in 25 mM piperazine-HCl buffer (pH 5.0). The fractions were applied to a Q XL column (GE Healthcare). The proteins were then eluted using 60 ml of 25 mM phosphate buffer, 60 ml of a linear gradient to 0.2 M NaCl, 60 ml of 0.5 M NaCl, and 60 ml of 1 M NaCl. The fractions with N-acetylation activity eluted by 0.5 M NaCl were collected and applied to a Mono Q HR 5/5 column (GE Healthcare). The proteins were eluted with a linear gradient of 0 to 1 M NaCl.

Protein identification by mass spectrometry.

Proteins from the Mono Q column were separated on an SDS-PAGE gel (18). The bands were excised, destained with 25 mM NH₄HCO₃–50% acetonitrile, cut into smaller pieces, and dried in vacuo. Proteins were then digested with trypsin (Promega, Madison, WI)–25 mM NH₄HCO₃ (pH 8.3) at 37°C for 16 h. The resulting peptides were extracted with aqueous 70% acetonitrile–5% formic acid by the use of sonication and lyophilized.

The tryptic peptides from each gel band were analyzed by reversed-phase nanoflow liquid chromatography–tandem mass spectrometry (LC-MS/MS) (19). Briefly, the peptides were redissolved in 0.1% formic acid and injected onto a fused silica capillary electrospray ionization (ESI) C₁₈ column (9 cm by 75 μm [inside diameter]), which was coupled online to an Orbitrap mass spectrometer (LTQ-Orbitrap XL; Thermo Electron, San Jose, CA). Mobile phases A (0.1% formic acid–water) and B (0.1% formic acid–acetonitrile) were delivered by a Dionex UltiMate 3000 Nano and Cap LC system (Dionex Softron GmbH, Germering, Germany). Peptides were separated using a step gradient of 2% to 42% solvent B for 40 min and 42% to 98% solvent B for 10 min at a flow rate of 250 nl/min. The mass spectrometer was operated in a data-dependent mode, in which the seven most abundant peptide molecular ions from an MS survey scan (acquired in the Orbitrap analyzer) were dynamically selected for collision-induced dissociation (CID) and analyzed in the linear ion trap using a normalized collision energy of 35%.

The raw data from tandem mass spectrometry were searched against the Microbacterium testaceum protein sequence database (20) of the National Center for Biotechnology Information (http://www.ncbi.nih.gov/). The SEQUEST cluster, running BioWorks (revision 3.3.1 SP1; Thermo Electron), was used for the identification of peptides and proteins.

Gene cloning.

Genomic DNA of Microbacterium sp. 4N2-2 was extracted with an UltraClean Microbial DNA isolation kit (MoBio Laboratories, Carlsbad, CA) following the manufacturer's instructions. Degenerate primers (5′-GGM CAG CTB TTC GAY GGM TCV TCS ATC CG-3′ and 5′-GSA GCT CGT AGA GGT CCT TGT CGA YSG GYK C-3′) were used for PCR. After the sequence of the amplicon had been obtained, two new primers were prepared to amplify upstream and downstream sequences to give the full sequence of the target gene. A random nonamer primer (GE Healthcare) and a reverse primer for upstream (5′-GTC ACG TCG GGG ATG AGC TGC ATG TCG-3′) were used to amplify the start region. A random nonamer primer and a forward primer for downstream (5′-TCG AAC CCG AAG GCC AAG CGC ATC GAG-3′) were used to amplify the stop region. Sequence fragments were assembled in silico to give the full nucleotide sequence.

Heterologous expression and purification of the protein.

The forward primer (5′-CCG TTC CTC CAG GAG TTG ACA TAT GTT CAC CAC C-3′) and reverse primer (5′-NNN NNN NGG ATC CTC ASA CSC CGW AGT ACA GCT-3′) for the pET-11a plasmid (Novagen) were designed to amplify the full target gene (glnA) from the genomic DNA of Microbacterium sp. 4N2-2 (underlined letters are recognition sequences for NdeI and BamHI, respectively). The sequence of the degenerate primer region was confirmed by comparing it with the full gene sequence obtained. The amplified target gene (glnA; 1,425 bp) was inserted at the NdeI and BamHI sites of pET-11a, and the expression construct was confirmed by restriction enzyme analysis and sequencing. The expression construct was then introduced into the E. coli BL21(DE3) pLysS expression host to yield E. coli BL21(DE3) pLysS/pET-11a-glnA. Expression of the target protein was controlled by the T7lac promoter of pET-11a by gradual addition of isopropyl-β-d-thiogalactopyranoside (IPTG) (final concentration, 0.8 mM) to cultures in LB medium for 1 h at 28°C.

After induction, the cultures were incubated for 4 h, harvested by centrifugation, washed twice with 10 mM phosphate buffer (pH 7.2), and centrifuged again. After a freeze-thaw step, they were resuspended in the same buffer. Cells from 1.5 liter were suspended in 25 ml of phosphate buffer and disrupted by sonication in an ethanol-ice bath. Cell debris was removed by centrifugation and filtration. Ammonium sulfate was gradually added, to a final concentration of 20% to 100%, to find the optimum concentration. The precipitate not containing the target protein, as shown by a norfloxacin N-acetylation assay and SDS-PAGE analysis, was pelleted and discarded.

The target protein was then precipitated from the supernatant with more ammonium sulfate. The pellet was resuspended in 10 mM Tris-HCl (pH 7.8) and dialyzed with the same buffer to 20 ml, using a Centricon centrifugal filter unit (EMD Millipore). A HiPrep Q XL 16/10 column (GE Healthcare) was used for purification with a gradient of 0 to 2.0 M NaCl in 10 mM Tris-HCl (pH 7.8) at a flow rate of 1 ml min⁻¹. The active fractions were concentrated and dialyzed with 20 mM Tris-HCl (100 mM NaCl, pH 7.5) to 5 ml. The fractions were applied in sequence to a gel filtration column (HiLoad 26/600 Superdex 200 prep grade; GE Healthcare). The target protein was eluted with 20 mM Tris-HCl (100 mM NaCl, pH 7.5) at a flow rate of 2 ml min⁻¹. Thyroglobulin (669 kDa), ferritin (440 kDa), aldolase (158 kDa), and conalbumin (75 kDa) from GE Healthcare were used as standards.

The eluted protein was concentrated and dialyzed in 50 mM HEPES buffer (pH 7.5) with 10% glycerol. Purification was monitored by SDS-PAGE analysis, and protein was analyzed by using a MicroBCA protein assay kit (Thermo Scientific).

Norfloxacin resistance as shown in broth cultures and a disk assay.

Overnight cultures of E. coli BL21(DE3) pLysS/pET-11a-glnA in LB medium were transferred into 200 μl of fresh LB medium as a 1% inoculum. Cells were grown at 28°C in 96-well plates with agitation in a microplate reader (Synergy 2; BioTek, Winooski, VT) at 28°C. When the optical density at 600 nm had reached 0.2, IPTG was added at a final concentration of 0.8 mM and norfloxacin was added at 0, 0.5, 1.0, 2.0, 5.0, 10.0, and 20.0 μg ml⁻¹. Growth was monitored every 15 min up to 16 h. All tests were repeated in triplicate.

For the disk assay, cells cultured overnight were spread with a sterilized cotton swab on LB agar, with or without 0.8 mM IPTG. A norfloxacin disk (Remel, Lenexa, KS) (10 μg) was applied to the middle of each plate. After 15 h at 28°C, the diameter of the clear zone was measured. All tests were repeated in triplicate.

Enzyme activity analysis and inhibitor screening.

Enzyme activity was analyzed with 0.06 mM substrate–0.2 mM acetyl-CoA–5 mM CaCl₂–50 μl of 50 mM Tris-HCl buffer–20 mM NaCl (pH 8.0). One enzyme unit was defined as the amount of enzyme that converts 1 μmol of norfloxacin to N-acetylnorfloxacin, with saturated acetyl-CoA, at pH 8.0 and 45°C in 1 h.

To screen potential inhibitors of N-acetylation activity, glutamate, glutamine, NH₄Cl, MgCl₂, MnCl₂, ATP, and ADP were prepared as 100 mM stock solutions and added to the reaction mixture at final 2 mM concentrations. Activity of the enzyme was expressed as percent residual activity compared to that of control reaction mixtures without inhibitors.

In silico docking analysis.

A three-dimensional (3-D) protein structure of the gene, based on the structure of the glutamine synthetase (GS) from Mycobacterium tuberculosis (21), was generated using the SWISS-MODEL server (22). PDBsum (http://www.ebi.ac.uk/pdbsum/) provided a quick overview, including schematic diagrams and interaction of GS with ligands (ADP, norfloxacin, and acetyl-CoA). Automated docking of norfloxacin and acetyl-CoA to GS was performed using AutoDock Vina (23). The grid box (Center_x = −25.368, Center_y = 83.074, Center_z = 11.188) was centered in the catalytic active region (size_x = 58, size_y = 50, and size_z = 56). The maximum number of binding modes to generate was 1,000 for all docking analyses. An input pdbqt file of GS, with norfloxacin having the lowest binding energy of −8.0 kcal mol⁻¹, was used for the AutoDock docking acetyl-CoA simulation. Initial attempts to choose the functional binding modes of norfloxacin and acetyl-CoA were guided by (i) the nucleotide ADP binding site and (ii) the distance (<5 Å) between the N-terminal of the piperazine ring and the S1P of acetyl-CoA. PyMOL (0.99RC6) (http://www.pymol.org/) was used to visualize all structural figures.

RESULTS

Partial purification and identification of protein responsible for norfloxacin N-acetylation.

The N-acetylation activity in Microbacterium sp. 4N2-2 increased for up to ∼7 to 9 days in OM medium with Casamino Acids (Fig. 1), so cell extracts were prepared from cultures grown in this medium for 9 days. Acetyl-CoA (14) was found to be required as an acetyl group donor for N-acetylation by cell extracts, and CaCl₂ was found to be required as a cofactor.

Fig 1.

Production of N-acetylnorfloxacin by Microbacterium sp. 4N2-2. The nitrogen sources were Casamino Acids (■), glutamate (□), glutamine (▲), and ammonium chloride (△). d, days.

To identify the responsible protein, selected bands containing partially purified protein (∼50 to 55 kDa) were excised from an SDS-PAGE gel and analyzed by MS/MS. Eight proteins were considered to be possibly responsible for N-acetylation of norfloxacin, but glutamine synthetase (GS) was considered the most likely because N-acetylation activity in Microbacterium sp. 4N2-2 is governed by the nitrogen source and is strongly enhanced by Casamino Acids (16). Expression of GS is also enhanced by Casamino Acids in Bacillus subtilis (24). Casamino Acids lacks glutamine (the product of GS, which causes feedback inhibition), but glutamate (the substrate) represents 8.4% to 15.9% of the total amino nitrogen (BD Bionutrients Technical Manual; BD Science).

To understand the relation between GS and norfloxacin N-acetylation, cultures of Microbacterium sp. 4N2-2 were grown with various nitrogen sources. Production of N-acetylnorfloxacin was found in media with Casamino Acids or glutamate but not in media containing glutamine or ammonium (Fig. 1).

Cloning of gene encoding N-acetyltransferase.

The gene encoding GS (glnA) was cloned for heterologous expression and then characterized biochemically to reveal any unexpected functions of the protein. In the absence of a genome database for Microbacterium sp. 4N2-2, degenerate PCR primers were used for cloning. Genes encoding GS (with the open reading frame [ORF] numbers from annotated genomes) from Microbacterium testaceum StLB037 (MTES3453), Clavibacter michiganensis subsp. sepedonicus (CMS1619), Leifsonia xyli CTCB07 (LXX10080), Arthrobacter phenanthrenivorans sphe3 (Asphe15760), Renibacterium salmoninarum ATCC 33209 (RSal2447), and Mycobacterium smegmatis mc²155 (MSMEG4290) were aligned and used for designing degenerate primers. The forward primer (nucleotides 145 to 173 of MTES3453) and the reverse primer (nucleotides 1,183 to 1,212 of MTES3453) were used to generate an amplicon (1.1 kbp) that was used subsequently to obtain sequence information for the middle region of glnA. Two other primers (a reverse primer for upstream and a forward primer for downstream), designed from the sequence information obtained, were used to amplify regions outward from the middle fragment of glnA. The fragments were assembled manually to give a full nucleotide sequence for glnA of Microbacterium sp. 4N2-2.

The cloned gene sequence, which was deposited in GenBank (JX901058), shared 86%, 82%, 81%, 73%, 73%, and 65% deduced amino acid sequence identities, respectively, with the strains used for primer design, whereas GS from E. coli K-12 shared only 54% identity with the deduced protein. Two other GSs of M. testaceum StLB037 (MTES3446 and MTES1058) shared 36% and 30% identities, respectively, with the deduced protein.

Comparison of MS/MS data with the sequence of GS.

After the cloning of glnA, peptide information for the GS of Microbacterium sp. 4N2-2 was sought in the MS/MS spectra measured previously during partial purification of the proteins. One suggested peptide sequence, based on the M. testaceum StLB037 genome, was IPITGSNPK. The corresponding deduced amino acid sequence of GS of Microbacterium sp. 4N2-2 was IPLTGSNPK; the MS/MS spectrum of this peptide was exactly identified (P ≤ 0.001) in the data. This implied that GS from the cell extract of Microbacterium sp. 4N2-2 was most likely the protein responsible for N-acetylation.

Heterologous expression of GS and activity analysis.

Based on the sequence obtained, an expression construct (pET-11a-glnA) was prepared and expressed in E. coli BL21(DE3) pLysS. Cells from IPTG-induced and noninduced cultures of E. coli, containing either pET-11a-glnA or pET-11a, were harvested and disrupted. GS, which was expressed only in the IPTG-induced cells with pET-11a-glnA, had the predicted size of ∼53 kDa (Fig. 2A). Cell extracts from these cultures showed high activity for converting norfloxacin to N-acetylnorfloxacin, whereas the low activity (background) in cell extracts from the controls presumably originated from E. coli enzymes.

Fig 2.

SDS-PAGE gels showing (A) heterologous expression of Microbacterium sp. 4N2-2 glutamine synthetase in the soluble fraction of E. coli BL21(DE3) pLysS/pET-11a-glnA with IPTG induction and (B) purification of glutamine synthetase of Microbacterium sp. 4N2-2 from heterologous overexpression.

Contribution of GS to norfloxacin resistance.

Because Microbacterium sp. 4N2-2 grew well with a norfloxacin concentration of more than 70 μg ml⁻¹, the contribution of GS to norfloxacin resistance was not evaluated directly in this strain. A norfloxacin-sensitive strain, E. coli BL21(DE3) pLysS, was used to show the role of this enzyme in norfloxacin resistance. Under the conditions used for this assay, the enzyme was active in the soluble fraction.

In broth cultures of E. coli BL21(DE3) pLysS pET-11a-glnA without IPTG, 0.5 μg ml⁻¹ of norfloxacin inhibited growth (Fig. 3). Induced cells expressing heterologous GS did not grow as well as noninduced cells in the absence of norfloxacin, indicating that there was a metabolic load due to the heterologously expressed protein. After induction of GS, the induced cells grew in the presence of 0.5 and even 1.0 μg ml⁻¹ norfloxacin at a rate similar to that of induced cultures without norfloxacin. However, growth of both induced and noninduced cultures was inhibited by 2.0 μg ml⁻¹ norfloxacin (Fig. 3).

Fig 3.

Decreased sensitivity to norfloxacin (0 to 2 μg ml⁻¹) due to induction of glutamine synthetase in E. coli BL21(DE3) pLysS/pET-11a-glnA. IPTG-induced (■) and noninduced (□) liquid cultures were used to monitor norfloxacin sensitivity by induction of glutamine synthetase in the E. coli BL21(DE3) pLysS host.

The norfloxacin disk diffusion assay also supported the idea that GS contributed to norfloxacin resistance. The diameter of the inhibition zone for E. coli cells without glnA was not significantly altered by IPTG, but E. coli cells with glnA showed an ∼3-mm-smaller inhibition zone diameter (statistically significant; P < 0.05) in IPTG-induced cultures than in noninduced cultures (Fig. 4).

Fig 4.

Norfloxacin disk assay of noninduced (gray bars) and IPTG-induced (white bars) cultures of E. coli BL21(DE3) pLysS containing plasmids with or without the glnA insert encoding glutamine synthetase. The assay was used to monitor changes in norfloxacin sensitivity due to expression of glutamine synthetase. In the cells with the glnA insert, the reduced diameter of the clear zones of inhibition due to induction of glutamine synthetase was statistically significant (P < 0.05).

Purification of heterologously expressed GS.

GS was precipitated with 55% ammonium sulfate and then sequentially purified with an anion exchange column (Table 1). The molecular mass of GS, calculated as 53.32 kDa from its deduced amino acid sequence, coincided with the approximate size shown by SDS-PAGE (Fig. 2B). Gel filtration indicated a protein complex with an approximate size of 640 kDa, implying that, like other GSs (25), this GS may form a homododecamer.

Table 1.

Purification of recombinant glutamine synthetase from E. coli^a

Step	Total protein (mg)	Total activity (mU)	Specific activity (mU/mg)	Yield (%)
Crude cell extract	206.9	85.9	0.42	100.0
Ammonium sulfate precipitation	35.0	73.4	2.10	85.4
Anion exchange	3.1	7.0	2.26	8.2

^aActivities were determined with acetyl-CoA and CaCl₂.

Characterization of GS.

Using GS purified by ammonium sulfate precipitation and anion-exchange chromatography, optimum conditions for norfloxacin N-acetylation were evaluated. The optimum pH and temperature were determined to be 8.0 and 37 to 55°C, respectively. Acetyl-CoA at 0.2 mM was required as an acetyl group donor and calcium ion at 5.0 mM for enzyme activation. With 0.2 mM acetyl-CoA, there was ∼5% abiotic background in total production of N-acetylnorfloxacin at pH 8.0. Apparent K_m and V_max values were obtained from Lineweaver-Burk plots; with saturated acetyl-CoA, the K_m was 2,208 μM and V_max was 13.2 mU (mg protein)⁻¹.

Glutamate, ammonia, and glutamine did not inhibit norfloxacin N-acetylation by the purified enzyme, but ATP and ADP caused 48% and 39% inhibition, respectively (Fig. 5). This indicated that norfloxacin does not bind to the pocket for substrates and products but instead binds to the ATP and ADP binding site (21, 26). Mg²⁺ and Mn²⁺, which may activate GS for its usual substrates and coordinate the substrates at the active site (21, 27), instead caused 48% and 54% inhibition, respectively (Fig. 5).

Fig 5.

Inhibition of N-acetylation activity of purified glutamine synthetase from Microbacterium sp. 4N2-2 by substrates and cofactors (2 mM each).

In silico structural analysis.

The 3-D structural model of GS from Microbacterium sp. strain 4N2-2 (Fig. 6A) was constructed using the crystal structure (Protein Data Bank code 2BVC) of the GS from Mycobacterium tuberculosis (21) as a template. The root mean square (RMS) deviation (Cα) of the GS model (474 amino acids) superimposed on the 2BVC structure (486 amino acids) was 0.79 Å, and 471 amino acid residues were aligned with 63.4% sequence identity. Figure 6A shows a biologically relevant dodecamer model with two hexameric rings.

Fig 6.

Model of Microbacterium sp. 4N2-2 glutamine synthetase structure and potential binding modes of norfloxacin and acetyl-CoA. (A) Top and side views of glutamine synthetase complex. The dimeric active site is indicated by a white square on the surface representation. (B and C) Electrostatic surface diagrams of the substrate binding pocket, colored by charge, and the potential binding modes of norfloxacin (B; five potential positions) and acetyl-CoA (C; 10 potential positions).

In GS, the active site between subunits has a cleft ∼14 Å in length, ∼9 Å in width, and ∼11 Å in height. In the GS-ADP complex, ADP is placed in almost the same manner in the cleft, as shown in the 2BVC structure (Fig. 6B). The AutoDock docking norfloxacin-GS simulation proposed 504 possible binding solutions. About 16 of them were located at the binding site of ADP, in which the reaction center (-N) of the piperazine ring lies appropriately in the cleft. Figure 6B shows the five binding modes with the lowest binding energy (∼−8.00 kcal mol⁻¹), all of which have similar conformations and locations in a cleft of GS.

Figure 6C shows 10 potential functional binding modes of acetyl-CoA with the lowest binding energies of ∼−6.00 kcal mol⁻¹ from 687 possible binding solutions. As revealed in a closeup of the binding modes of acetyl-CoA (Fig. 6C), despite its dispersed binding positions, most of the functional acetyl group moieties resided in the bottom of a cleft bound to the piperazine ring of norfloxacin.

DISCUSSION

Many bacterial enzymes that N-acetylate various substrates use acetyl-CoA as an acetyl donor (28–30). In bacteria as well as humans, N-acetylation may be involved in detoxification of drugs (28, 29, 31, 32), including fluoroquinolones (14). The activity of GS in members of the Actinomycetales is regulated by transcriptional regulation and posttranslational adenylylation (33–35). The peptide sequence of GS was found in MS/MS spectra from the partial isolation of proteins of Microbacterium sp. 4N2-2, and it exactly matched the deduced protein sequence for GS. When it was heterologously expressed in E. coli and purified, GS also showed norfloxacin N-acetylation activity.

Although N-acetylation of antibiotics by GS has not been previously reported, there are reports of N-acetylation of proteins by GS. A so-called “moon lighting property” of GS in Mycobacterium spp. is the N-acetylation of glutathione S-transferase, with a polyphenolic acetate as an acetyl group donor (36, 37). The measured activity of Microbacterium sp. 4N2-2 GS for norfloxacin N-acetylation was not as high as that for glutamine synthesis; neither was it as high as the activities of other N-acetyltransferases (14, 36, 37), as shown by the K_m and V_max values. The K_m for norfloxacin was similar to the K_m for glutamate, but it was more than 10-fold higher than the K_m values reported for ATP and NH₄Cl with the GS of Nocardia asteroides (38). High K_m and low V_max values may be explained by an accidental substrate-enzyme relationship. Thus, the contribution of GS to resistance, which is effective only for low norfloxacin concentrations, may be understood from the high K_m and low V_max.

Together with the biochemical observations, including (i) the low enzyme activity, (ii) acetyl-CoA as an acetyl group donor, (iii) the relatively high K_m for norfloxacin, (iv) the inhibition by ATP and ADP, and (v) the Ca²⁺ requirement, the structural model also provides insights into the N-acetylation activity of GS. This enzyme not only requires acetyl-CoA and Ca²⁺ for N-acetylation of norfloxacin, but it also has a structural environment suitable for functional binding of the substrates. When the predicted binding was examined with respect to the active site of the enzyme, the geometry of the active site seemed unlikely to be problematic for binding. Moreover, with norfloxacin bound in the active site, the piperazine ring would orient toward the open space, which simultaneously could accept the acetyl group of acetyl-CoA. As revealed in the docking analysis of the complex, when GS bound norfloxacin with the lowest binding energy (−8.0 kcal mol⁻¹), most of the acetyl group of the bound acetyl-CoA was located in the bottom of a cleft, placing the substrate in an ideal position for functional group donation. The substrate binding modes and geometry of the active site support the idea that GS can join norfloxacin and acetyl-CoA in the active site. The N-terminal of the piperazine ring lies near the reactive S1P atom of acetyl-CoA, which is an appropriate position for functional group transfer. Therefore, the relatively low N-acetylation activity of GS suggests that the reaction depends on the precisely coordinated functional binding of the substrates, norfloxacin and acetyl-CoA, and Ca²⁺.

The GS of Microbacterium sp. 4N2-2 appears to belong to a family of highly conserved bacterial GSs (39). Other bacterial GSs that acetylate fluoroquinolones may also exist, considering that GS is conserved in actinobacteria with high structural homology (21, 39) and that N-acetylation activity for norfloxacin is found in several Mycobacterium spp. (15). This idea also is supported by the protein N-acetylation function of GS of M. smegmatis (37).

The study of various GSs for N-acetylation activity, and enzyme engineering based on models of protein-ligand binding, should provide insight into the N-acetylation function of GS. Moreover, because N-acetylation is an important detoxification mechanism for bioactive molecules (40, 41), GSs of environmental bacteria may be able to modify various chemicals, including antibiotics.

In nutrient-limited and biofilm bacteria, the starvation response mediated by guanosine pentaphosphate is associated with tolerance to antibiotics, including fluoroquinolones (42). Expression of GSs of E. coli, Bacillus subtilis, and Mycobacterium spp. is also enhanced by the stringent response of nutrient starvation, especially with respect to ammonium (43–45). This implies that in oligotrophic sites or biofilms, a bacterial strain with a GS similar to Microbacterium sp. 4N2-2 might show fluoroquinolone resistance. Most environmental sites have low-level (nanogram per liter) contamination by fluoroquinolones (46–48), so activation of GS by nutrient nitrogen limitation could contribute to fluoroquinolone resistance.

This report describes the modification of a fluoroquinolone by an enzyme from an environmental bacterium. The approach used in this research, including selective isolation of the bacterium, enzyme activity analysis, protein isolation and characterization, and structure modeling, represents a useful strategy for enzyme identification and characterization in studies of biotransformation and antibiotic resistance.