ATP Exhibits Antimicrobial Action by Inhibiting Bacterial Utilization of Ferric Ions

ATP up-regulates macrophage antimycobacterial activity in a P2X7-dependent manner, but little is known about whether ATP directly exhibits antimicrobial effects against intracellular mycobacteria. In this study, we found that ATP inhibited the growth of various bacteria, including Staphylococcus, Pseudomonas, and mycobacteria, without damaging bacterial surface structures. Using gene technology, we newly established an enterobactin-deficient (entB−) mutant from ATP-resistant Klebsiella pneumoniae, and found the recovery of ATP susceptibility in the enterobactin-deleted mutant. Therefore, ATP's antibacterial activity is attributable to its iron-chelating ability. Since ATP distributed in the cytosol of macrophages at high concentrations, ATP appears to augment macrophage's antimicrobial activity by directly attacking intracytosolic and intra-autophagosomal pathogens. Furthermore, ATP exhibited combined effects with some antimicrobials against methicillin-resistant S. aureus (MRSA) and M. intracellulare, suggesting its usefulness as an adjunctive drug in the chemotherapy of certain intractable infections.

resistant to ATP (Supplemental Fig. S1e-g). ATP susceptibility of Pseudomonas aeruginosa markedly varied from strain to strain ( Fig. 1g; Supplemental Fig. S1h). These findings indicate the species and strain dependency of common bacteria for their ATP susceptibility 12 . In separate experiments, we determined the effective concentrations of ATP which cause 50% inhibition of the growth of the two ATP-susceptible bacterial species with a serial dilution method, as follows: 0.60 mM for P. aeruginosa strain P2 and 0.77 mM for M. intracellulare strain N-260.
Antimicrobial effect of ATP correlates with metal-ion chelating activity. ATP chelates metal ions, such as Mg 21 and Mn 21 , causing modulation of various enzymatic reactions in vivo. Therefore, ATP may exhibit antimicrobial effects on certain microorganisms by depriving them of essential metal ions. This aspect was confirmed as follows. First, ATP's anti-M. intracellulare activity was significantly reduced by MgCl 2 and FeCl 3 (Fig. 2a). Second, all test metal chelators exhibited antimicrobial activity against M. intracellulare in the order of EDTA . EGTA . ATP . pyrophosphate (iron-chelator), and the excess free Mg 21 ion-occupying chelating ability of these agents partly blocked their antibacterial effects (Fig. 2b). Indeed, it has been demonstrated that EDTA exhibits strong antimicrobial activity against P. aeruginosa, and the antibacterial activity of certain antibiotics which inhibit bacterial protein synthesis was reported to be significantly potentiated when used in combination with EDTA, presumably due to the chelation of Ca 21 and Mg 21 ions 13 .
Antimicrobial effects of these chelators were markedly decreased when ATP low-susceptible M. avium was used as a target (Supplemental Fig. S3a). In common bacteria, a good correlation was also observed between their susceptibility to ATP and chelators ( Fig. 2c; Supplemental Fig. S3b, c). Therefore, ATP's antimicrobial effect is attributable to its metal-ion chelating action.
Ferric ion is essential for the growth and survival of most living organisms 12,14,15 . During infection in host animals, bacterial organisms are obligated to compete with host iron-chelating substances, such as transferrin and lactoferrin, to acquire ferric ions. By using siderophores, bacterial high-affinity ferric chelators, bacteria maintain intracellular iron concentration at sufficient levels for their survival and growth in hosts 12,14,15 . As shown in Fig. 2d, ATP exhibited significant levels of ferric ion-chelating activity. In this context, we compared the profiles of the antibacterial actions of ATP with EDTA using M. intracellulare and P. aeruginosa as target organisms. ATP www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 8610 | DOI: 10.1038/srep08610 (5 mM) exhibited levels of growth-inhibiting activity against these bacteria comparable to those of EDTA (0.5 mM) (Fig. 2e, f). The combined use of FeCl 3 (ferric ions) (0.5 mM) markedly reduced the antibacterial activity of ATP as well as EDTA, strongly suggesting that the antibacterial activities of ATP and EDTA are principally based on their ferric ion-chelating ability. Next, we found that ATP-resistant E. coli and K. pneumoniae, but not ATP-susceptible S. aureus, produced significant levels of siderophores when cultivated in the presence of 2 or 5 mM ATP (Fig. 2g-i). In P. aeruginosa, siderophore production was observed for only ATP-resistant strains (Fig. 2j). Therefore, the ATP-induced production of siderophores seems to be limited to ATP-resistant bacteria.
Siderophore interferes with ATP's antimicrobial effects. Next, to confirm that the antibacterial activity of ATP is mediated by its ironchelating effects, we established an enterobactin-deficient mutant strain of K. pneumoniae by deleting the entB gene encoding 2,3-dihydro-2,3-dihydrooxybenzoate synthase, a key enzyme for enterobactin biosynthesis (Fig. 3a) 16 . A PCR experiment confirmed that this strain has a complete deficit in the entB locus (Fig. 3a), resulting in a marked reduction in its siderophore-producing ability (Fig. 3b). While siderophore production by the parent ATPresistant strain was augmented in the presence of ATP or another iron chelator 2,29-dipyridyl, such a phenomenon was not observed in the entB gene-KO mutant strain. This suggests that the major siderophore, the production of which is up-regulated by ATPmediated iron starvation, is enterobactin (Fig. 3c). Notably, ATP exhibited moderate levels of antimicrobial effects on the entB 2 mutant strain, but not on the entB 1 parent strain producing enterobactin (Fig. 3d). Therefore, it is concluded that the antimicrobial activity of ATP is mediated by its iron-chelating ability.
ATP directly acts on intramacrophage bacteria in the cytosol. Next, we investigated the cellular mechanisms of ATP's antimicrobial action, as follows. First, ATP caused the early-phase inhibition of macromolecule synthesis (protein, RNA, and DNA) by S. aureus, but the inhibition thereafter diminished, presumably due to genetic adaptation to iron starvation (Fig. 4a) 17 . Second, ATP exhibited no bactericidal effect on M. intracellulare (Supplemental Fig. S4a) and caused no damage to the bacterial surface structure, including cell www.nature.com/scientificreports membranes, in terms of the release of intracellular proteins (Supplemental Fig. S4b). This aspect is supported by electron microscopy of the surface structure of ATP-treated M. intracellulare, indicating no obvious damage to the cell wall and cytoplasmic membrane (Fig. 4b). Third, intracellular concentrations of ATP in various cells including MWs are reported to be 1 to 8 mM 18,19 . Therefore, MWs are capable of acquiring sufficient intracellular ATP concentration to exhibit antimicrobial effects.
Based on this, we carried out some experiments to confirm the relationship between MW antibacterial activity and the intramacrophage ATP concentration, and obtained the following findings. Firstly, J774.1 MWs were measured for changes in the intracellular ATP concentration during the course of M. intracellulare infection. Interestingly, the intracellular ATP concentration was maintained at around 1 mM for up to 4 h after bacterial infection, showing that the intracellular paucity of ATP did not occur in MWs because of bacterial invasion (Fig. 4c). Importantly, this finding indicates that, in MWs engulfing bacteria, sufficient intracellular ATP concentrations for the efficacious expression of ATP antibacterial activity are maintained during the course of infection. In addition, pretreatment of MWs with a representative MW-activating cytokine, IFN-c, with or without subsequent triggering with phorbol myristate acetate (PMA), did not cause a significant increase in the intracellular ATP concentration (Fig. 4d). This finding suggests that the ATP's antimicrobial action against intramacrophage bacteria is not potentiated in connection with MW activation elicited by IFN-c-priming and PMA-triggering. It thus appears that intracytosolic ATP only partially contributes to the manifestation of the antibacterial activity of MW as a basic intramacrophage antimicrobial system. Notably, PMA-triggering caused a marked decrease in the intramacrophage ATP concentration, suggesting the rapid consumption of intracytosolic ATP, presumably due to the PMA-induced potentiation of intracellular metabolic pathways related to inflammatory reactions elicited by PMA-mediated signaling.
Since MW apoptotic events lead to the elevation of cytosolic ATP concentrations, ATP may also contribute to the apoptosis-mediated induction of MW antimycobacterial activity 19 . Thus, ATP is expected to participate in the expression of MW's antimycobacterial function. Interestingly, ATP exhibited antagonistic effects on the bactericidal activity of reactive nitrogen intermediates and reactive oxygen intermediates, the major intraphagosomal antimicrobial effector molecules of MWs (Fig. 4e), while it exhibited additive effects in combination with cationic antimicrobial peptides (Fig. 4f). Next, we examined whether or not ATP is co-localized in MW phagosomes engulfing bacteria. ATP was recovered from top and bottom fractions corresponding to the cytosol and possibly also in the lysosomal vesicles (Fig. 4g). Although M. intracellulare-engulfing phagosomes were also sedimented at bottom position, its peak fraction (Fr. 28) did not correspond to the position of ATP sedimentation. Taken together, ATP seems not to actively participate in intraphagosomal killing of M. intracellulare, but plays important roles in the growth inhibition of the organisms residing in the cytosol of MWs.
ATP enhances the activity of antimicrobial drugs for clinical treatment of intractable infections. Iron acquisition in the host milieu is critical for the expression of virulence by Vibrio vulnificus, M. tuberculosis, and fungi. Thus, iron overload in hosts occasionally causes the progression and deterioration of infection due to these pathogens 12,14,15 . Hence, trials of iron-chelating therapy have been carried out 12,20 . Here, we examined the profile of ATP's antimicrobial  activity from the viewpoint of its application to the treatment of bacterial infections. Notably, ATP augmented the antimicrobial activity of vancomycin against MRSA (Fig. 4h). ATP also potentiated the antimicrobial activity of clarithromycin/rifampin and clarithromycin/ethambutol regimens against Mycobacterium avium complex (MAC) (Fig. 4i). Thus, it is promising to use ATP as an adjunctive agent in chemotherapy for intractable MRSA and MAC infections. Indeed, ATP is widely distributed within the animal body, is essentially non-toxic, and exhibits no severe side effects on the human body.

Discussion
In this study, ATP was found to exhibit growth inhibitory activity against some kinds of bacteria, including S. aureus, P. aeruginosa, and slowly growing mycobacteria. In this context, it has been reported that mammalian siderophore, 2,5-dihydroxy benzoic acid, promotes the growth of E. coli under the Fe 31 -depleted condition and consequently causes the reduction of host resistance to E. coli infection 21 . Interestingly, ATP having iron-chelating activity behaved in a manner strikingly opposite to the mammalian siderophore. Notably, ATP does not facilitate but interferes with bacterial iron uptake.
Important findings concerning the biological mechanisms of ATP's antimicrobial effects are as follows. ATP exhibited ferric ion-chelating activity and ATP's antimicrobial activity was mimicked by a ferric ion-chelating agent pyrophosphate and metal ion chelators, EDTA and EGTA. Indeed, the occupation of ferric ionbinding activity of ATP with MgCl 2 and FeCl 3 blocked ATP's antimicrobial action. Chrome Azurol S (CAS) assay indicated that  ATP-resistant bacterial organisms, but not ATP-susceptible organisms, produced ferric ion chelators. Importantly, ATP-resistant K. pneumoniae strain acquired ATP susceptibility by losing its ability to produce enterobactin due to deletion of the entB gene. These findings support the very novel view that ATP exhibits antimicrobial effects by depriving ferric ions from target bacteria through its iron-chelating effects.
With special reference to the biological significance of ATP's antimicrobial effects, it is of interest to clarify whether ATP plays significant roles in the antimicrobial mechanisms of host MWs. Among MW organelles, ATP was specifically localized in the cytosol and possibly also in the lysosomal vesicles, as previously reported 22,23 , but not in M. intracellulare-engulfing phagosomes (Fig. 4g). Notably, the anti-M. intracellulare activity of reactive nitrogen and oxygen intermediates, the most important antimicrobial effectors of MWs, was completely blocked by ATP (Fig. 4e), although ATP potentiated the anti-M. intracellulare activity of the cationic antimicrobial peptide cathepsin G, a lysosomal antimicrobial serine protease (Fig. 4f). Taken together, we can exclude the possibility that ATP plays central roles in intraphagosomal inhibition and the killing of infected mycobacteria inside MWs. It is possible that ATP plays important roles in the inhibition of bacterial growth in the cytosol of MWs. In this context, it is generally considered that the intramacrophage survival and growth of virulent mycobacteria occur mainly due to the inhibition of phagosomal maturation and phagosomelysosome fusion by mycobacterial virulence factors, including lipoarabinomannan, protein kinase G, and protein phosphatase A 24,25 . Notably, van der Wel et al. reported the opposite finding that phagosome-lysosome fusion rapidly occurs even in MWs engulfing virulent mycobacteria including M. tuberculosis 26 . They found that live M. tuberculosis translocates from phagolysosomes into cytosol, and cytosolic entry is dependent on bacterial secretion of ESAT-6 and CFP-10 proteins 26 . Indeed, after phagocytic internalization into host MWs, M. tuberculosis and MAC escape from phagosomes into cytosol with the aid of tlyA gene-encoded cytolysin 26,27 . Thus, bacterial evasion of MW's intraphagosomal killing system is an important strategy of virulent mycobacteria to survive inside host MWs. The present findings suggest that ATP acts as an inhibitor of the intracytoplasmic growth of bacteria, which evade MW phagosomes after infecting host MWs via phagocytosis.
In this context, autophagosome-mediated bacterial inhibition may be important for the elimination of intracytosolic mycobacteria, which evade intraphagosomal killing. Indeed, autophagy plays an important role in host immunological defenses against pathogenic mycobacteria 28 . Therefore, there is a possibility that autophagic events in infected MWs may be connected with ATP-mediated upregulation of the antimycobacterial function of host MWs. IFNc-induced autophagy, causing the maturation of mycobacteriacontaining autophagosomes to autophagolysosomes, is effective in the elimination of mycobacteria that have escaped from phagosomes into cytosol 29 . Furthermore, ATP induces autophagy in MWs and causes rapid bacterial killing within autophagosomes 30 . Therefore, it is likely that cytosolic ATP is accumulated in autophagosomes together with mycobacterial organisms and exhibits antimicrobial action against bacteria residing within autophagolysosomes. This view should be confirmed by systematic experiments on the autophagosome-mediated killing of intracytosolic bacteria, and further studies are currently underway.  F-1, F-19, F-20), and common bacteria including S. aureus (strain 209P, Smith, S2, S3, S4, S5, S6, S7), L. monocytogenes strain EGD, and E. coli (strain K-12, 81, E1, E2, E3, E4, E5, E6), K. pneumoniae (strain I0004, I0008, I0029, I0036, I0058, SMU-1), and P. aeruginosa (strain P1, P2, P3, P4, P5, P6). Test mycobacteria and common bacteria were grown in Middlebrook 7H9 medium and Tryptic soy broth, respectively. Mice. Twelve-week-old male BALB/c mice purchased from Japan Clea Co., Osaka, Japan, were used. Animal care and experimental procedures were approved by the Animal Research Committee of Shimane University and conducted according to the Regulations for Animal Experimentation at Shimane University.

Organisms
Antimicrobial activity of test agent against extracellular organisms. The activities of test agents against extracellular mycobacteria were measured as follows. Test organisms (10 5 CFUs) were suspended in 0.5 ml 7HSF medium, a broth medium with the same composition as 7H11 agar without malachite green, containing test agents, and cultivated at 37uC in a CO 2 incubator (5% CO 2 -95% humidified air). After 4-d cultivation, the incubation mixture was centrifuged at 1,000 3 g for 10 min, and the recovered micro-organisms were then washed with distilled water by centrifugation. The resulting bacterial pellet was suspended in 0.5 ml distilled water and the CFUs of recovered organisms were counted on the 7H11 agar plate. Next, in cases of common bacteria, test microorganisms were cultivated in Tryptic soy broth containing the test agents at 37uC for 3 to 10 h. The CFUs in the resultant culture were counted on heart infusion agar plates.
Measurement of bactericidal activities of MW's antimicrobial effectors. M. intracellulare strain N-260 was incubated in 0.1 M sodium acetate buffer (pH 5.5) with or without the addition of prescribed concentrations of (1) cathepsin G, adefensin, NaNO 2 , or H 2 O 2 1 NaI 1 FeSO 4 at the cell density of 1 3 10 6 CFU/ml at 37uC. After individual treatment of test microorganisms for 2 h or 4 d, residual bacterial CFUs were counted on 7H11 agar.
Establishment of enterobactin-deficient mutant strain of K. pneumoniae. An enterobactin-deficient mutant strain of K. pneumoniae was established by deleting the entB gene (Fig. 3a). Briefly, plasmid DNA was constructed by conjugating a negative-selection suicide vector containing the sacB gene and the K. pneumoniae enterobactin locus, from which the entB gene was deleted (Fig. 3a: upper illustration). The plasmid was transfected into E. coli strain S-17-1, and the resultant transformant was co-cultivated with K. pneumoniae strain SMU-1 to allow recombination between the transfected enterobactin locus and the same DNA region of the parent K. pneumoniae strain, providing an entB deletion mutant. PCR experiment confirmed that this strain has a complete deficit in the entB locus (Fig. 3a). The detailed experimental procedures are as follows. The whole genome DNA of K. pneumoniae strain SMU-1 (wild-type strain) was prepared using Instagene matrix. To generate a deletion mutant of the entB gene, upstream (1518 bp) and downstream (1133 bp) regions of its ORF were amplified using two PCR primer sets: (1) entB-F1-Sal (forward primer), 59-cagctggtcgacaaccttgcctg-39, and entB-R1-Nh (reverse primer), 59-ttcgccgggctagcgctggcttc-39, and (2) entB-F2-Nh (forward primer), 59aggctagccgtttgccaccgaaac-39, and entB-R2-RI (reverse primer), 59agcaggggaattcccggatggcgg-39. The resultant DNA amplicons were subjected to enzymatic digestion with Sal I and Nhe I endonucleases and with Nhe I and EcoR I endonucleases, and the obtained DNA fragments were then integrated into a pK18mobSacB plasmid DNA between its Sal I and Eco RI restriction sites. After electroporation-based incorporation of the resultant plasmid DNA into E. coli strain S17-1, the plasmid was further transferred to K. pneumoniae strain SMU-1 via bacterial conjugation, allowing homologous recombination between the plasmid DNA and the corresponding genetic region of the host genome DNA. Then, bacteria having incorporated plasmids were deleted by performing cultivation on an LB agar plate containing 10% sucrose. The deletion of the entB gene in the recombinant K. pneumoniae was confirmed by genomic PCR testing using a primer set: entB-F3 (forward primer), 59-ccgaaggcaagctgggctgaaggag-39, and entB-R3 (reverse primer), 59-gcatcggcgacgtccagcgtttcgg-39. Measurement of bacterial siderophore production by the CAS assay. Bacterial siderophore production was measured by the CAS assay, as follows. Test bacteria were cultured in LB medium with or without the addition of prescribed concentrations of ATP or dipyridyl at 37uC for 1 h. The resultant bacteria were then collected by centrifugation, re-suspended in modified M9 minimal medium, and cultivated at 37uC for 48 h. The resultant culture fluid was mixed with an equal volume of CAS reagent, and the OD 620 of the resultant mixture was measured after 15-min incubation at room temperature. Siderophore production was represented by the decrease in the OD 620 value from that of the control incubation mixture without the addition of bacteria.
Measurement of macromolecule biosynthesis by S. aureus. S. aureus strain 209P was cultured in the presence or absence of ATP at a cell density of 5 3 10 5 CFU/ml in 0.2 ml Tryptic soy broth containing [ 14 C] isoleucine (1 mCi/ml) or [ 14 C] uracil (0.1 mCi/ml) for the measurement of protein and DNA/RNA synthesis, respectively, in a microculture well for up to 20 h. For the measurement of protein and whole nucleic acid (DNA plus RNA) synthesis, the bacterial culture was added to 1% bovine serum albumin, the bacterial culture was spotted onto paper filter (Whatmann, Maldstone, England). The paper filter was thereafter rinsed twice with 10% trichloroacetic acid solution and with ethanol once, dried at 60uC, immersed in toluene-based scintillant, and counted for radioactivity using Tri-Carb 2100TR (Perkin Elmer, Inc., Waltham, MA). For the measurement of whole DNA synthesis, the bacterial culture was pretreated with 1 M KOH for 24 h and subjected to the same procedures as for protein and whole nucleic acid measurement, as described above.
Measurement of ATP concentrations in MW's organelles and cytosol. J774.1 mouse MWs (4 3 10 7 ), which had been pretreated with interferon-c (50 U/ml) for 24 h, were cultivated at the cell density of 8 3 10 6 cells/dish in 10% fetal bovine serum-RPMI 1640 medium in the presence of 4 3 10 8 CFU/dish FLUOS-labeled M. intracellulare strain N-260 (MOI 5 50) at 37uC for 2 h. Infected MWs were harvested by centrifugation, suspended in 1 ml of 0.25 M sucrose buffer containing protease inhibitors (PMSF, leupeptin, pepstatin A, and EDTA), and homogenized using a Dounce homogenizer (Wheaton Science Products, Millville, NJ). After 5-min centrifugation at 200 3 g, the resultant supernatant was subjected to Percoll-gradient ultracentrifugation. Briefly, the supernatant was mixed with 9 volumes of 18% Percoll and centrifuged at 36,000 3 g for 1 h. After centrifugation, the Percoll gradient formed in the centrifuge tube was fractionated into 30 fractions, and each fraction was measured for FLUOS fluorescence intensity (excitation at 488 nm and emission at 535 nm) and its ATP concentration was determined by the usual luciferase assay. In addition, b-hexosaminidase (lysosomes) and lactate dehydrogenase (cytosol) were measured in the gradient fractions by the usual methods.
Effects of ATP on cell-surface structures of mycobacteria. ATP-mediated impairment of the bacterial cell membrane was measured in terms of the release of cellular proteins, as follows. M. intracellulare strain N-260 was incubated in phosphate-buffered saline (PBS) with or without the addition of 5 mM ATP at 37uC for up to 96 h. At intervals, the bacterial culture was harvested, centrifuged at 5,000 3 g for 10 min, and the resultant supernatant was measured for the concentration of cellular proteins released from bacterial cell bodies by the Bradford assay. Furthermore, bacterial surface structures were observed directly using transmission electron microscopy. According to the usual method, test bacterial cells were treated with glutaraldehyde-based fixative solution, embedded in epoxy resin, and polymerized at 60uC. For ultrastructural studies, thin sections were stained with uranyl acid and lead citrate prior to examination using a transmission electron microscope at a JEM-1200EX (JEOL, Tokyo, Japan).