Biochemical Characterization of Quinolinic Acid Phosphoribosyltransferase from Mycobacterium tuberculosis H37Rv and Inhibition of Its Activity by Pyrazinamide

Quinolinic acid phosphoribosyltransferase (QAPRTase, EC 2.4.2.19) is a key enzyme in the de novo pathway of nicotinamide adenine dinucleotide (NAD) biosynthesis and a target for the development of new anti-tuberculosis drugs. QAPRTase catalyzes the synthesis of nicotinic acid mononucleotide from quinolinic acid (QA) and 5-phosphoribosyl-1-pyrophosphate (PRPP) through a phosphoribosyl transfer reaction followed by decarboxylation. The crystal structure of QAPRTase from Mycobacterium tuberculosis H37Rv (MtQAPRTase) has been determined; however, a detailed functional analysis of MtQAPRTase has not been published. Here, we analyzed the enzymatic activities of MtQAPRTase and determined the effect on catalysis of the anti-tuberculosis drug pyrazinamide (PZA). The optimum temperature and pH for MtQAPRTase activity were 60°C and pH 9.2. MtQAPRTase required bivalent metal ions and its activity was highest in the presence of Mg2+. Kinetic analyses revealed that the Km values for QA and PRPP were 0.08 and 0.39 mM, respectively, and the kcat values for QA and PRPP were 0.12 and 0.14 [s-1], respectively. When the amino acid residues of MtQAPRTase, which may interact with QA, were substituted with alanine residues, catalytic activity was undetectable. Further, PZA, which is an anti-tuberculosis drug and a structural analog of QA, markedly inhibited the catalytic activity of MtQAPRTase. The structure of PZA may provide the basis for the design of new inhibitors of MtQAPRTase. These findings provide new insights into the catalytic properties of MtQAPRTase.


Introduction
Tuberculosis (TB) is a chronic infectious disease, caused by the intracellular pathogen Mycobacterium tuberculosis, with an estimated 8.7 million cases and 1.4 million deaths each year according to a 2012 World Health Organization (WHO) Report [1]. The emergence of resistance to anti-TB drugs, in particular multidrug-resistant TB (MDR-TB), is a public health problem and poses a serious threat to global control of TB [2][3][4]. Therefore, there is an urgent need for new countermeasures against TB. To address this issue, the aim of the present study was to define the functions of poorly characterized enzymes that may provide targets for designing new drugs to eradicate M. tuberculosis infections.
Quinolinic acid phosphoribosyltransferase (QAPRTase; EC 2. 4.2.19) is encoded by nadC and is a key enzyme in the de novo pathway of nicotinamide adenine dinucleotide (NAD) biosynthesis [5][6][7]. NAD is a coenzyme of pivotal importance in maintaining redox balance and energy metabolism and is continuously interconverted between oxidized (NAD) and reduced (NADH) forms [8]. In most bacteria, NAD biosynthesis is essential for cell survival and viability [9], which makes it an attractive target for the development of new antibacterial drugs, with steps shared by de novo and recycling pathways as a source of candidate enzymes for therapeutic intervention [5,[10][11][12].
QAPRTase catalyzes the Mg 2+ -dependent transfer of the phosphoribosyl moiety from 5-phosphoribosyl-1-pyrophosphate (PRPP) to the nitrogen atom of quinolinic acid (QA) to generate nicotinic acid mononucleotide (NAMN), pyrophosphate (PPi), and CO 2 (Fig. 1) [5,[13][14][15]. QA is the first intermediate in the de novo pathway of NAD biosynthesis that is common to all organisms and is mainly produced by the degradation of tryptophan in most eukaryotes [5,16,17]. In contrast, in prokaryotes, including M. tuberculosis, it is produced from L-aspartate and dihydroxyacetone phosphate by the enzymes encoded by nadA (quinolinic acid synthetase) and nadB (L-aspartate oxidase) [18,19]. In M. tuberculosis, nadA, nadB, and nadC are encoded in a single operon (nadABC) which is regulated by a repressor encoded by nadR [5,20].
QAPRTase is a type II PRTase that participates in the de novo pathway of the pyridine coenzyme NAD [7,15]. Recently, nicotinic acid phosphoribosyltransferase (NAPRTase) and nicotinamide phosphoribosyltransferase, which are involved in the salvage pathways of NAD biosynthesis, have been classified as type II PRTases [15,23,24,27]. The activities of QAPRTase and NAPRTase were similar, although they are specific for their respective substrates [28,29]. M. tuberculosis relies entirely on the de novo pathway of NAD for survival; therefore, it should be extremely vulnerable to drugs targeted against QAPRTase. The crystal structure of QAPRTase from M. tuberculosis (MtQAPRTase) is known [5]; however, the biochemical properties of MtQAPR-Tase remain to be determined. Therefore, in the present study, we examined and characterized the enzymatic activities of MtQAPR-Tase.
QA is a structural analog of the anti-tuberculosis prodrug pyrazinamide (PZA), and pyrazinoic acid (POA) is its active form. PZA is an important component of first line anti-TB drugs in the chemotherapy for TB and MDR-TB [30,31]. Mycobacteria acquire resistance to PZA through mutations in the gene encoding pyrazinamidase (PZase), an enzyme that converts PZA to the active anti-bacterial form of POA [30,32,33]. Although mutations in PZase (encoded by pncA) responsible for the generation of most PZA-resistant M. tuberculosis strains have been identified [9], some PZA-resistant M. tuberculosis strains do not harbor pncA mutations [33]. The mechanism of action and main target of PZA are still not clearly understood; however, intensive investigations are in progress across laboratories worldwide [30][31][32][33][34]. Recently, Shi W. et al. [31] reported that the PZA inhibits trans-translation in M.
tuberculosis, and suggest that POA binds to ribosomal protein S1 (RpsA) and subsequently inhibits trans-translation. Therefore, PZA may have interfered with bacterial growth and survival.
Because PZA and POA are structural analogs of QA, we reasoned that MtQAPRTase may use PZA or POA as a substrate, or these substrates may inhibit the enzymatic activity of MtQAPRTase.
In the present study, we examined and characterized the enzymatic activities of wild type (WT) plus mutant MtQAPRTases and the effect of PZA and POA on WT MtQAPRTase in vitro.

Bacterial strains and plasmid
Escherichia coli strain DH5a (Life Technologies) was used as the host for molecular cloning. E. coli strain BL21 (DE3) was purchased from Merck KGaA (Darmstadt, Germany) and used for protein expression. The pET-30a plasmid (Merck KGaA) was used construct in an expression vector to produce WT and mutant versions of recombinant MtQAPRTase.
Cloning and mutagenesis of nadC from M. tuberculosis H37Rv genomic DNA Genomic DNA from M. tuberculosis H37Rv was isolated as previously described [35,36]. The nadC (Rv1596, accession number; NP_216112.1) of M. tuberculosis H37Rv was amplified from genomic DNA [20] by using the polymerase chain reaction (PCR). The reaction mixture (20 mL) contained long and accurate (LA) PCR buffer II (Mg 2+ -free); 2.5 mM MgCl 2 ; 200 mM each of dATP, dCTP, dGTP, and dTTP; 250 ng of genomic DNA from M. tuberculosis H37Rv; 1.25 units of LA Taq DNA polymerase (all from TaKaRa Bio, Kyoto, Japan); and 0.1 mM of each primer. The primers are shown in Table 1. PCR was conducted using a Takara PCR Thermal Cycler Dice Mini (TaKaRa Bio Inc., Shiga, Japan) as follows: pre-denaturation at 98uC for 2 min, 35 cycles of denaturation at 98uC for 10 sec, annealing at 55uC for 10 sec and extension at 72uC for 2 min, and final extension at 72uC for 2 min. K-001 and K-003 primers were used to amplify WT nadC (Table 1). Nucleotide sequences encoding a 6x-histidine residue cluster were added directly upstream of the nadC stop codon on Cterminal. The PCR product (885 bp) was ligated to the TA cloning plasmid and used to transform E. coli DH5a. The recombinant nadC plasmid DNA was recovered from the colonies and digested with NdeI and XhoI. The nadC product was ligated to pET-30a expression vector that was digested with the same restriction endonucleases. Mutant nadC genes were generated from WT nadC by using a QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies, Inc., Santa Clara, CA) according to the manufacturer's instruction. The list of primers used to generate mutants is shown in Table 1. After mutagenesis, plasmids were recovered and purified using a Promega Minipreps DNA purification kit (Madison, WI, USA). WT and mutant plasmids were confirmed by sequencing with ABI Prism BigDye Termina-tor v3.1 Cycle Sequencing kit (Life Technologies). The sequencing reactions were performed according to the manufacturer's instructions. The sequencing products were analyzed using an ABI Prism 3130xl Genetic Analyzer (Life Technologies). The sequences generated by the software were compared with their respective nadC sequence using Bioedit software (http://www. bioedit.com/). Using the molecular modeling program MOE (Molecular Operating Environment, Chemical Computing Group, Montreal, Canada), we constructed a few models, which were elongated by six histidines at the C-terminus. These models were generated and structure energies were minimized using CHARMM27 force field (gradient below 0.5); the resulting structures were used as starting conformations for molecular dynamics simulations (MD) for 100 ps at 300 K. During the simulation, all amino acid residues were flexible.

Expression and purification of MtQAPRTase WT and mutants
WT and mutant forms of MtQAPRTase were purified as described previously [35,37] with the following modifications. E. coli BL21 (DE3) was transformed with expression vectors carrying the M. tuberculosis WT or mutant nadC. Single colonies were picked and grown overnight at 37uC in 4 mL of Luria-Bertani (LB) medium containing 50 mg/mL kanamycin. Overnight cultures (2 mL) were used to inoculate 200 mL of LB medium containing kanamycin. Cells were then cultured at 37uC for 5 h, until the optical density (OD) at 600 nm reached 0.8 to 1.0. Expression of recombinant enzymes was induced with 1 mM IPTG, followed by incubation at 14uC for 18 h. The bacteria were harvested by centrifugation at 13,0006g at 4uC for 10 min, the pellets were stored at 280uC for 12 h, then suspended in 7 mL of binding buffer (20 mM sodium phosphate [pH 7.4], 0.5 M NaCl, 40 mM imidazole), and disrupted by sonication at 80% pulsar power, 40 sec on/1 min off 10-times on ice, using a UP50H sonicator (Hielscher Ultrasonic, Teltow, Germany). The extracts were centrifuged at 13,0006g at 4uC for 10 min, and the supernatants were harvested. The cell extracts were injected onto a HisTrap HP column (1.662.5 cm; GE Healthcare Bio-Sciences, Buckinghamshire, UK) pre-equilibrated with deionized water and binding buffer. The columns were washed with binding buffer until the absorbance reached a steady baseline, and the proteins were eluted using an elution buffer containing 20 mM sodium phosphate (pH 7.4), 0.5 M NaCl, and 0.5 M imidazole. The

Kinetic parameters of MtQAPRTase
The kinetic parameters of recombinant MtQAPRTase activities were performed by a monitoring spectrophotometric assay using Ultrospec 3000 UV/Visible Spectrophotometer (GE Healthcare Biosciences, Buckinghamshire, UK). Typical assay mixtures contained 50 mM KH 2 PO 4 (pH 7.2), 6 mM MgCl 2 , various concentration of QA (0.02-0.4 mM), PRPP (0.02-1.5 mM), and 1.62 mM of purified WT MtQAPRTase in a total volume of 100 mL. A spectrophotometric assay of QAPRTase activity over 20 min at 37uC measured the increase in absorbance at 266 nm resulting from the conversion of QA to NAMN (DA266 = 920 M cm 21 ) [16,39]. Ultrospec 3000 UV/Visible Spectrophotometer plotted the increase in absorbance at 266 nm (dA) against time (min) and calculated the initial slope (dA/min) automatically. The initial velocity was calculated from the slope (dA/min). The reaction mixtures for determination of K m values from initial velocity data were prepared using various concentrations of PRPP and a fixed concentration of QA (0.3 mM); conversely, reaction mixtures were also prepared using various concentrations of QA and a fixed concentration of PRPP (1.0 mM). The reaction was initiated by the addition of WT MtQAPRTase, and absorbance was immediately estimated. To determine K m and k cat values when using QA and PRPP as substrates, their concentrations were varied. Kinetic parameters for MtQAPRTase activity were calculated using GraphPad Prism 5 software (GraphPad Software,

Molecular docking study
The molecular docking and visualization studies were performed using MF mypresto v1.2 (Fiatlux Corporation, Tokyo, Japan) molecular modeling software. Coordinates of MtQAPR-Tase for structure-based molecular modeling were retrieved from the Protein Data Bank (PDB), USA (http://www.rcsb.org/pdb/), under the accession code [PDB ID: 1QPQ] [5]. The molecular docking models used dimeric forms of A and B subunits from the three-dimer form of MtQAPRTase and the coordinates of ligands and waters were manually removed. On the basis of the information regarding the QA-binding site, the location and size of the receptor pocket were set (Center: X: -15.796 Y: 42.865 Z: 17.445, and radius: 4 angstrom). Optional parameters in MF mypresto v1.2 were used to create a topology file, which included addition of hydrogen atoms, calculation of a grid potential, and docking simulation. Flexible docking method was used and the scores are expressed as a sum of five potentials: accessible surface area, coulomb potential, hydrogen bonds, hydrogen bond considering anisotropy, and van der Waals interactions. The protein-ligand binding free energies were estimated by MF mypresto v1.2. The results of molecular docking were visualized, and the distance between residues of amino acid and PZA was calculated using PyMOL v1.3 (http://www.pymol.org/) and WinCoot-0.7.2 (http://www.ysbl.york.ac.uk/).

Expression and purification of MtQAPRTase
Full-length nadC from M. tuberculosis H37Rv was inserted into the expression vector pET-30a downstream of the T7 promoter to express a His-tagged recombinant MtQAPRTase. We confirmed  the association between the location of His-tag and active site of MtQAPRTase by using MOE. This analysis shows that the Histag could be located outside the active site of MtQAPRTase (Fig.  S1), suggesting that the His-tag does not interfere with the activity of MtQAPRTase. Furthermore, it was confirmed that His-tag could not affect the multimerization of MtQAPRTase (Fig. S1). DNA sequence analysis of the recombinant plasmid confirmed the identity and integrity of nadC and verified that no mutations were introduced during PCR amplification. Recombinant MtQAPR-Tase was purified to homogeneity using a two-step column chromatographic procedure described in Materials and Methods. The molecular mass of recombinant MtQAPRTase determined using SDS-PAGE was 31 kDa (Fig. 2A, lane 2), consistent with that calculated from the amino acid sequence containing six histidine residues (30773.9 Da). The specific activity of purified recombinant MtQAPRTase was 1.2 U/mg with a yield of 19% and purification of 6.4-fold ( Table 2). The molecular mass of MtQAPRTase as estimated using gel-filtration chromatography column was approximately 58 kDa (Fig. 2B), indicating that the enzyme exists as a dimer in solution.

Enzymatic activities of MtQAPRTase
The enzymatic activities of MtQAPRTase were determined using HPLC (Fig. 3). QA and NAMN were observed at 6.0 and 12.5 min, respectively (Fig. 3A), while a peak of PRPP was not detected using our separation conditions. When MtQAPRTase was added to a reaction mixture, the area of the QA and NAMN peaks decreased and increased, respectively (Fig. 3B). We then confirmed that recombinant MtQAPRTase converts QA and PRPP to NAMN, PP i , and CO 2 in the presence of Mg 2+ .
The optimum temperature of MtQAPRTase was 60uC, and its activity was decreased at 70uC (Fig. 4A). Its pH optimum was 9.2, and its activity could not be detected at pH values below 5.2 and above 11.0 (Fig. 4B).
QAPRTase isolated from M. tuberculosis, E. coli, and Salmonella typhimurium requires Mg 2+ for activity [12,15,18]. Similarly, MtQAPRTase activity requires bivalent metal ions such as Mg 2+ . Activity was highest in the presence of Mg 2+ ( Table 3). The relative activities of MtQAPRTase in the presence of 6 mM metal ions are shown in Table 3.

Kinetic parameters of WT MtQAPRTase
The Michaelis-Menten plot was used to estimate K m and k cat according to the activity at different concentrations of PRPP

Site-directed mutagenesis and activities of mutant MtQAPRTases
X-ray crystallographic studies of QAPRTase indicate that Arg1059, Arg139, Arg162, Lys172, and His175 are located in the QA-binding site of MtQAPRTase (Fig. 6) [5,[13][14][15]40]. Arg1059 is present in another subunit. To determine whether these residues play an important role in the enzymatic activity of MtQAPRTase, we constructed, expressed, and purified the following MtQAPR-Tase-single mutants: Arg1059Ala, Arg139Ala, Arg162Ala, Ly-s172Ala, and His175Ala ( Fig. 2A and 6). The catalytic activities of the MtQAPRTase mutants could not be detected when purified preparations were evaluated at variable concentrations up to 38.9 mM, which was more than 24-fold higher than normal condition (1.62 mM). Therefore, these results indicate that these

Inhibitory effects of MtQAPRTase activity by PZA/POA and IC 50 values
Because PZA and POA are structural analogs of QA (Fig. 7A), we predicted that these molecules may be substrates or inhibitors of MtQAPRTase. When PZA or POA were added to reaction mixtures instead of QA or PRPP, no new peaks were detected (data not shown), indicating that they are not substrates. To determine the inhibitory effects of PZA and POA on MtQAPR-Tase activity, PZA or POA was incubated with MtQAPRTase under different pH conditions (pH 7.2 and 6.2). Initially, we investigated the concentrations of PZA and POA under neutral (pH 7.2) and weak acidic (pH 6.2) conditions. PZA and POA were tested for their abilities to inhibit MtQAPRTase between the concentrations of 0.1 and 3.0 mM (data not shown). Figure 7B shows the results of a representative inhibitory QAPRTase assay containing 1 mM of PZA or POA at pH 7.2 (Fig. 7B left) and pH 6.2 (Fig. 7B right), respectively. Interestingly, the inhibitory effect of PZA at pH 7.2 condition was dramatically higher than at pH 6.2 condition and that of POA at any pH value. IC 50 values of PZA and POA are summarized in Fig. 7C

Molecular docking study
According to the values of the protein-ligand binding free energy and the score, the optimum docking model was selected. When the PZA was used as the ligand in molecular docking, the values of the protein-ligand binding free energy and the score were 25.22 and 21.80, respectively. On the other hand, when the QA was used as the ligand, the values of the protein-ligand binding free energy and the score were 25.87 and 22.07, respectively. There was no significant difference between the values of the proteinligand binding free energy and the score of PZA and QA. Molecular docking results are shown in Fig. 8. Although the structure of QAPRTase has been reported as a three-dimer form [5,16], we used dimeric forms of A and B subunits for the molecular docking study. Our docking results showed that PZA binds to Arg139, His161, Arg162, and Ser248 through a hydrogen-bonding network ( Fig. 8A and Table 4). Arg139 and Arg162 are involved in QA binding too (Fig. 6) [5]. Therefore, we suggest that the PZA-binding site of MtQAPRTase overlaps the QA-binding site (Fig. 8B). Arg139, His161, Arg162, and Ser248 were changed to Ala after PZA was docked to MtQAPRTase by using the WinCoot-0.7.2 program. We found that Ala139, Ala161, Ala162, and Ala248 did not interact with PZA (data not shown); thus, we believe that these amino acid residues play an important role in the interaction between MtQAPRTase and PZA (Fig. 8)

Discussion
We describe here the molecular cloning, expression, and purification of MtQAPRTase to determine its biochemical properties. MtQAPRTase forms a dimer in solution, which is consistent with its crystal structure [5]. The optimum temperature and pH of MtQAPRTase were 60uC and pH 9.2 and are similar to those of the QAPRTase from Alcaligenes eutrophus nov. subsp. Quinolinicus [42]. Because the enzymatic activity of MtQAPRTase was detected at pH 7.2 and 37uC (Fig. 4), we suggest that MtQAPRTase functions as a QAPRTase in M. tuberculosis.
Liu H. et al. [16] reported the kinetic characterization of QAPRTase (hQAPRTase) from Homo sapiens. We compared the kinetic studies of MtQAPRTase with those of hQAPRTase and recalculated the values of k cat from V max values. The k cat value for QA determined using the MtQAPRTase assay was 0.12 [s 21 ] and that reported for hQAPRTase was 0.04 [s 21 ]. The K m values for QA when evaluating MtQAPRTase and hQAPRTase were 80 and 22 mM, respectively. Therefore, we suggest that the enzymatic activity of MtQAPRTase is similar to that of hQAPRTase.
The inhibitory effects of PZA or POA on MtQAPRTase activity were determined at neutral (7.2) and weak acidic (6.2) pH, because it was reported that in M. tuberculosis, at an acidic external pH. In M. tuberculosis, at an acidic external pH, the rate of passive transmembrane equilibrium of POA apparently overwhelms that of active efflux, resulting in a huge accumulation of POA in the cells [31,43]. Although the internal pH in M. tuberculosis exposed to an acidic external pH is not known, we speculate that the internal pH is weakly acidic. Therefore, we evaluated the inhibitory effects at pH 6.2. Our results show that the inhibitory effect of PZA at neutral pH condition was dramatically higher than the inhibitory effect of POA at any pH value (Fig. 7B). Therefore, we suggest that PZA may inhibit the MtQAPRTase activity at neutral pH.
Structural studies indicate that Arg1059, Arg139, Arg162, Lys172, and His175 residues interact with QA, suggesting that the C3 carboxylate group of QA forms hydrogen bonds with the side-chain atoms of Arg162 and Arg139, whereas the C2 carboxylate group is within hydrogen-bonding distance of the main chain of Arg139 and the side chains of Arg1059 and Lys172 (Fig. 6) [5]. Further, the side chain of His161 is within van der Waals distance of the substrate [5]. We found that the four mutated amino acids play an important role in enzymatic activity of MtQAPRTase. QA binds tightly to QA-binding site (Arg1059, Arg139, Arg162, and Lys172) through hydrogen bonds (H-N-H_O-H and H-N_O-H) and other bonds (Fig. 6) [5], but not to alanine mutants (Ala1059, Ala139, Ala162, and Ala172) because of the absence of hydrogen-bonding (data not shown). The subsequent loss of hydrogen bonding could induce a substantial conformational change, which disrupts enzymatic activity of MtQAPRTase.
We further demonstrated that PZA strongly inhibited MtQAPRTase activity (Fig. 7). Based on the crystal structure and data on the interactions between QA and QAPRTase [5,13], we hypothesize that the PZA-binding site of QAPRTase overlaps the QA-binding site, because the structures of PZA and QA are very similar (Fig. 7A). Therefore, we investigated the binding mode of a PZA instead of a QA within the active site of MtQAPRTase using molecular docking analysis (Fig. 8). Our molecular docking results show that the PZA binds to amino acid residues (Arg139, His161, Arg162, and Ser248) through a hydrogen-bonding network ( Fig. 8A and Table 4), and QA binds to amino acid residues (Arg1059, Arg139, Arg162, and Lys172) on MtQAPRTase (Fig. 8B) [5]. We suggest further that Arg139 and Arg162 are essential amino acid residues for the formation of QAand PZA-binding sites, and His161 and Ser248 may be important only for recognition of PZA.
In conclusion, the present study identifies PZA as an inhibitor of MtQAPRTase. Base on the structure of PZA new inhibitors of MtQAPRTase could be designed. Because MtQAPRTase is a potential candidate target of new anti-TB drugs [5,[10][11][12], we expect that our present findings will contribute to their development.